2023-01-05 20:25:55 +08:00
|
|
|
/**************************************************************************/
|
|
|
|
/* runtime_interop.cpp */
|
|
|
|
/**************************************************************************/
|
|
|
|
/* This file is part of: */
|
|
|
|
/* GODOT ENGINE */
|
|
|
|
/* https://godotengine.org */
|
|
|
|
/**************************************************************************/
|
|
|
|
/* Copyright (c) 2014-present Godot Engine contributors (see AUTHORS.md). */
|
|
|
|
/* Copyright (c) 2007-2014 Juan Linietsky, Ariel Manzur. */
|
|
|
|
/* */
|
|
|
|
/* Permission is hereby granted, free of charge, to any person obtaining */
|
|
|
|
/* a copy of this software and associated documentation files (the */
|
|
|
|
/* "Software"), to deal in the Software without restriction, including */
|
|
|
|
/* without limitation the rights to use, copy, modify, merge, publish, */
|
|
|
|
/* distribute, sublicense, and/or sell copies of the Software, and to */
|
|
|
|
/* permit persons to whom the Software is furnished to do so, subject to */
|
|
|
|
/* the following conditions: */
|
|
|
|
/* */
|
|
|
|
/* The above copyright notice and this permission notice shall be */
|
|
|
|
/* included in all copies or substantial portions of the Software. */
|
|
|
|
/* */
|
|
|
|
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
|
|
|
|
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
|
|
|
|
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. */
|
|
|
|
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
|
|
|
|
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
|
|
|
|
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
|
|
|
|
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
|
|
|
|
/**************************************************************************/
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
#include "runtime_interop.h"
|
|
|
|
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
#include "core/config/engine.h"
|
2022-02-28 04:57:39 +08:00
|
|
|
#include "core/debugger/engine_debugger.h"
|
|
|
|
#include "core/debugger/script_debugger.h"
|
2021-09-13 01:50:13 +08:00
|
|
|
#include "core/io/marshalls.h"
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
#include "core/object/class_db.h"
|
|
|
|
#include "core/object/method_bind.h"
|
2021-09-13 01:50:13 +08:00
|
|
|
#include "core/os/os.h"
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
#include "core/string/string_name.h"
|
|
|
|
|
|
|
|
#include "../interop_types.h"
|
|
|
|
|
2021-09-13 02:23:05 +08:00
|
|
|
#include "modules/mono/csharp_script.h"
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
#include "modules/mono/managed_callable.h"
|
2021-09-13 02:23:05 +08:00
|
|
|
#include "modules/mono/mono_gd/gd_mono_cache.h"
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
#include "modules/mono/signal_awaiter_utils.h"
|
|
|
|
|
|
|
|
#ifdef __cplusplus
|
|
|
|
extern "C" {
|
|
|
|
#endif
|
|
|
|
|
2021-09-13 01:49:39 +08:00
|
|
|
// For ArrayPrivate and DictionaryPrivate
|
|
|
|
static_assert(sizeof(SafeRefCount) == sizeof(uint32_t));
|
|
|
|
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
typedef Object *(*godotsharp_class_creation_func)();
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
MethodBind *godotsharp_method_bind_get_method(const StringName *p_classname, const StringName *p_methodname) {
|
2021-12-29 06:25:16 +08:00
|
|
|
return ClassDB::get_method(*p_classname, *p_methodname);
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godotsharp_class_creation_func godotsharp_get_class_constructor(const StringName *p_classname) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
ClassDB::ClassInfo *class_info = ClassDB::classes.getptr(*p_classname);
|
|
|
|
if (class_info) {
|
|
|
|
return class_info->creation_func;
|
|
|
|
}
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
Object *godotsharp_engine_get_singleton(const String *p_name) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return Engine::get_singleton()->get_singleton_object(*p_name);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_stack_info_vector_resize(
|
2022-02-28 04:57:39 +08:00
|
|
|
Vector<ScriptLanguage::StackInfo> *p_stack_info_vector, int p_size) {
|
|
|
|
return (int32_t)p_stack_info_vector->resize(p_size);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_stack_info_vector_destroy(
|
2022-02-28 04:57:39 +08:00
|
|
|
Vector<ScriptLanguage::StackInfo> *p_stack_info_vector) {
|
|
|
|
p_stack_info_vector->~Vector();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_script_debugger_send_error(const String *p_func,
|
2022-02-28 04:57:39 +08:00
|
|
|
const String *p_file, int32_t p_line, const String *p_err, const String *p_descr,
|
|
|
|
bool p_warning, const Vector<ScriptLanguage::StackInfo> *p_stack_info_vector) {
|
|
|
|
EngineDebugger::get_script_debugger()->send_error(*p_func, *p_file, p_line, *p_err, *p_descr,
|
|
|
|
true, p_warning ? ERR_HANDLER_WARNING : ERR_HANDLER_ERROR, *p_stack_info_vector);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_internal_script_debugger_is_active() {
|
2022-02-28 04:57:39 +08:00
|
|
|
return EngineDebugger::is_active();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
GCHandleIntPtr godotsharp_internal_object_get_associated_gchandle(Object *p_ptr) {
|
2021-09-13 02:23:05 +08:00
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
CRASH_COND(p_ptr == nullptr);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (p_ptr->get_script_instance()) {
|
|
|
|
CSharpInstance *cs_instance = CAST_CSHARP_INSTANCE(p_ptr->get_script_instance());
|
|
|
|
if (cs_instance) {
|
|
|
|
if (!cs_instance->is_destructing_script_instance()) {
|
2022-02-28 04:57:30 +08:00
|
|
|
return cs_instance->get_gchandle_intptr();
|
|
|
|
}
|
|
|
|
return { nullptr };
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void *data = CSharpLanguage::get_existing_instance_binding(p_ptr);
|
|
|
|
|
|
|
|
if (data) {
|
|
|
|
CSharpScriptBinding &script_binding = ((RBMap<Object *, CSharpScriptBinding>::Element *)data)->get();
|
|
|
|
if (script_binding.inited) {
|
|
|
|
MonoGCHandleData &gchandle = script_binding.gchandle;
|
|
|
|
return !gchandle.is_released() ? gchandle.get_intptr() : GCHandleIntPtr{ nullptr };
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return { nullptr };
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_object_disposed(Object *p_ptr, GCHandleIntPtr p_gchandle_to_free) {
|
2022-02-28 04:57:30 +08:00
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
CRASH_COND(p_ptr == nullptr);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (p_ptr->get_script_instance()) {
|
|
|
|
CSharpInstance *cs_instance = CAST_CSHARP_INSTANCE(p_ptr->get_script_instance());
|
|
|
|
if (cs_instance) {
|
|
|
|
if (!cs_instance->is_destructing_script_instance()) {
|
|
|
|
cs_instance->mono_object_disposed(p_gchandle_to_free);
|
2021-09-13 02:23:05 +08:00
|
|
|
p_ptr->set_script_instance(nullptr);
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void *data = CSharpLanguage::get_existing_instance_binding(p_ptr);
|
|
|
|
|
|
|
|
if (data) {
|
|
|
|
CSharpScriptBinding &script_binding = ((RBMap<Object *, CSharpScriptBinding>::Element *)data)->get();
|
|
|
|
if (script_binding.inited) {
|
2022-02-28 04:57:30 +08:00
|
|
|
if (!script_binding.gchandle.is_released()) {
|
|
|
|
CSharpLanguage::release_binding_gchandle_thread_safe(p_gchandle_to_free, script_binding);
|
2021-09-13 02:23:05 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_refcounted_disposed(Object *p_ptr, GCHandleIntPtr p_gchandle_to_free, bool p_is_finalizer) {
|
2021-09-13 02:23:05 +08:00
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
CRASH_COND(p_ptr == nullptr);
|
|
|
|
// This is only called with RefCounted derived classes
|
|
|
|
CRASH_COND(!Object::cast_to<RefCounted>(p_ptr));
|
|
|
|
#endif
|
|
|
|
|
|
|
|
RefCounted *rc = static_cast<RefCounted *>(p_ptr);
|
|
|
|
|
|
|
|
if (rc->get_script_instance()) {
|
|
|
|
CSharpInstance *cs_instance = CAST_CSHARP_INSTANCE(rc->get_script_instance());
|
|
|
|
if (cs_instance) {
|
|
|
|
if (!cs_instance->is_destructing_script_instance()) {
|
|
|
|
bool delete_owner;
|
|
|
|
bool remove_script_instance;
|
|
|
|
|
2022-02-28 04:57:30 +08:00
|
|
|
cs_instance->mono_object_disposed_baseref(p_gchandle_to_free, p_is_finalizer,
|
|
|
|
delete_owner, remove_script_instance);
|
2021-09-13 02:23:05 +08:00
|
|
|
|
|
|
|
if (delete_owner) {
|
|
|
|
memdelete(rc);
|
|
|
|
} else if (remove_script_instance) {
|
|
|
|
rc->set_script_instance(nullptr);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Unsafe refcount decrement. The managed instance also counts as a reference.
|
|
|
|
// See: CSharpLanguage::alloc_instance_binding_data(Object *p_object)
|
|
|
|
CSharpLanguage::get_singleton()->pre_unsafe_unreference(rc);
|
|
|
|
if (rc->unreference()) {
|
|
|
|
memdelete(rc);
|
|
|
|
} else {
|
|
|
|
void *data = CSharpLanguage::get_existing_instance_binding(rc);
|
|
|
|
|
|
|
|
if (data) {
|
|
|
|
CSharpScriptBinding &script_binding = ((RBMap<Object *, CSharpScriptBinding>::Element *)data)->get();
|
|
|
|
if (script_binding.inited) {
|
2022-02-28 04:57:30 +08:00
|
|
|
if (!script_binding.gchandle.is_released()) {
|
|
|
|
CSharpLanguage::release_binding_gchandle_thread_safe(p_gchandle_to_free, script_binding);
|
2021-09-13 02:23:05 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_internal_signal_awaiter_connect(Object *p_source, StringName *p_signal, Object *p_target, GCHandleIntPtr p_awaiter_handle_ptr) {
|
2021-09-13 02:23:05 +08:00
|
|
|
StringName signal = p_signal ? *p_signal : StringName();
|
|
|
|
return (int32_t)gd_mono_connect_signal_awaiter(p_source, signal, p_target, p_awaiter_handle_ptr);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
GCHandleIntPtr godotsharp_internal_unmanaged_get_script_instance_managed(Object *p_unmanaged, bool *r_has_cs_script_instance) {
|
2021-09-13 02:23:05 +08:00
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
CRASH_COND(!p_unmanaged);
|
|
|
|
CRASH_COND(!r_has_cs_script_instance);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (p_unmanaged->get_script_instance()) {
|
|
|
|
CSharpInstance *cs_instance = CAST_CSHARP_INSTANCE(p_unmanaged->get_script_instance());
|
|
|
|
|
|
|
|
if (cs_instance) {
|
|
|
|
*r_has_cs_script_instance = true;
|
|
|
|
return cs_instance->get_gchandle_intptr();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
*r_has_cs_script_instance = false;
|
2022-02-28 04:57:30 +08:00
|
|
|
return { nullptr };
|
2021-09-13 02:23:05 +08:00
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
GCHandleIntPtr godotsharp_internal_unmanaged_get_instance_binding_managed(Object *p_unmanaged) {
|
2021-09-13 02:23:05 +08:00
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
CRASH_COND(!p_unmanaged);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
void *data = CSharpLanguage::get_instance_binding(p_unmanaged);
|
2022-02-28 04:57:30 +08:00
|
|
|
ERR_FAIL_NULL_V(data, { nullptr });
|
2021-09-13 02:23:05 +08:00
|
|
|
CSharpScriptBinding &script_binding = ((RBMap<Object *, CSharpScriptBinding>::Element *)data)->value();
|
2022-02-28 04:57:30 +08:00
|
|
|
ERR_FAIL_COND_V(!script_binding.inited, { nullptr });
|
2021-09-13 02:23:05 +08:00
|
|
|
|
|
|
|
return script_binding.gchandle.get_intptr();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
GCHandleIntPtr godotsharp_internal_unmanaged_instance_binding_create_managed(Object *p_unmanaged, GCHandleIntPtr p_old_gchandle) {
|
2021-09-13 02:23:05 +08:00
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
CRASH_COND(!p_unmanaged);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
void *data = CSharpLanguage::get_instance_binding(p_unmanaged);
|
2022-02-28 04:57:30 +08:00
|
|
|
ERR_FAIL_NULL_V(data, { nullptr });
|
2021-09-13 02:23:05 +08:00
|
|
|
CSharpScriptBinding &script_binding = ((RBMap<Object *, CSharpScriptBinding>::Element *)data)->value();
|
2022-02-28 04:57:30 +08:00
|
|
|
ERR_FAIL_COND_V(!script_binding.inited, { nullptr });
|
2021-09-13 02:23:05 +08:00
|
|
|
|
|
|
|
MonoGCHandleData &gchandle = script_binding.gchandle;
|
|
|
|
|
|
|
|
// TODO: Possible data race?
|
|
|
|
CRASH_COND(gchandle.get_intptr().value != p_old_gchandle.value);
|
|
|
|
|
|
|
|
CSharpLanguage::get_singleton()->release_script_gchandle(gchandle);
|
|
|
|
script_binding.inited = false;
|
|
|
|
|
|
|
|
// Create a new one
|
|
|
|
|
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
CRASH_COND(script_binding.type_name == StringName());
|
|
|
|
#endif
|
|
|
|
|
|
|
|
bool parent_is_object_class = ClassDB::is_parent_class(p_unmanaged->get_class_name(), script_binding.type_name);
|
2022-02-28 04:57:30 +08:00
|
|
|
ERR_FAIL_COND_V_MSG(!parent_is_object_class, { nullptr },
|
2021-09-13 02:23:05 +08:00
|
|
|
"Type inherits from native type '" + script_binding.type_name + "', so it can't be instantiated in object of type: '" + p_unmanaged->get_class() + "'.");
|
|
|
|
|
|
|
|
GCHandleIntPtr strong_gchandle =
|
|
|
|
GDMonoCache::managed_callbacks.ScriptManagerBridge_CreateManagedForGodotObjectBinding(
|
|
|
|
&script_binding.type_name, p_unmanaged);
|
|
|
|
|
2022-02-28 04:57:30 +08:00
|
|
|
ERR_FAIL_NULL_V(strong_gchandle.value, { nullptr });
|
2021-09-13 02:23:05 +08:00
|
|
|
|
|
|
|
gchandle = MonoGCHandleData(strong_gchandle, gdmono::GCHandleType::STRONG_HANDLE);
|
|
|
|
script_binding.inited = true;
|
|
|
|
|
|
|
|
// Tie managed to unmanaged
|
|
|
|
RefCounted *rc = Object::cast_to<RefCounted>(p_unmanaged);
|
|
|
|
|
|
|
|
if (rc) {
|
|
|
|
// Unsafe refcount increment. The managed instance also counts as a reference.
|
|
|
|
// This way if the unmanaged world has no references to our owner
|
|
|
|
// but the managed instance is alive, the refcount will be 1 instead of 0.
|
|
|
|
// See: godot_icall_RefCounted_Dtor(MonoObject *p_obj, Object *p_ptr)
|
|
|
|
rc->reference();
|
|
|
|
CSharpLanguage::get_singleton()->post_unsafe_reference(rc);
|
|
|
|
}
|
|
|
|
|
|
|
|
return gchandle.get_intptr();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_tie_native_managed_to_unmanaged(GCHandleIntPtr p_gchandle_intptr, Object *p_unmanaged, const StringName *p_native_name, bool p_ref_counted) {
|
2021-09-13 02:23:05 +08:00
|
|
|
CSharpLanguage::tie_native_managed_to_unmanaged(p_gchandle_intptr, p_unmanaged, p_native_name, p_ref_counted);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_tie_user_managed_to_unmanaged(GCHandleIntPtr p_gchandle_intptr, Object *p_unmanaged, Ref<CSharpScript> *p_script, bool p_ref_counted) {
|
2021-09-13 02:23:05 +08:00
|
|
|
CSharpLanguage::tie_user_managed_to_unmanaged(p_gchandle_intptr, p_unmanaged, p_script, p_ref_counted);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_tie_managed_to_unmanaged_with_pre_setup(GCHandleIntPtr p_gchandle_intptr, Object *p_unmanaged) {
|
2021-09-13 02:23:05 +08:00
|
|
|
CSharpLanguage::tie_managed_to_unmanaged_with_pre_setup(p_gchandle_intptr, p_unmanaged);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_new_csharp_script(Ref<CSharpScript> *r_dest) {
|
2022-02-28 04:57:36 +08:00
|
|
|
memnew_placement(r_dest, Ref<CSharpScript>(memnew(CSharpScript)));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_internal_script_load(const String *p_path, Ref<CSharpScript> *r_dest) {
|
2022-05-28 10:56:46 +08:00
|
|
|
Ref<Resource> res = ResourceLoader::load(*p_path);
|
|
|
|
if (res.is_valid()) {
|
|
|
|
memnew_placement(r_dest, Ref<CSharpScript>(res));
|
|
|
|
return true;
|
|
|
|
} else {
|
|
|
|
memnew_placement(r_dest, Ref<CSharpScript>());
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_internal_reload_registered_script(CSharpScript *p_script) {
|
2022-02-28 04:57:36 +08:00
|
|
|
CRASH_COND(!p_script);
|
|
|
|
CSharpScript::reload_registered_script(Ref<CSharpScript>(p_script));
|
2021-09-13 02:23:05 +08:00
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_filter_godot_objects_by_native(StringName *p_native_name, const Array *p_input, Array *r_output) {
|
2021-09-13 02:23:05 +08:00
|
|
|
memnew_placement(r_output, Array);
|
|
|
|
|
|
|
|
for (int i = 0; i < p_input->size(); ++i) {
|
|
|
|
if (ClassDB::is_parent_class(((Object *)(*p_input)[i])->get_class(), *p_native_name)) {
|
|
|
|
r_output->push_back(p_input[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_filter_godot_objects_by_non_native(const Array *p_input, Array *r_output) {
|
2021-09-13 02:23:05 +08:00
|
|
|
memnew_placement(r_output, Array);
|
|
|
|
|
|
|
|
for (int i = 0; i < p_input->size(); ++i) {
|
|
|
|
CSharpInstance *si = CAST_CSHARP_INSTANCE(((Object *)(*p_input)[i])->get_script_instance());
|
|
|
|
|
|
|
|
if (si != nullptr) {
|
|
|
|
r_output->push_back(p_input[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_ref_new_from_ref_counted_ptr(Ref<RefCounted> *r_dest, RefCounted *p_ref_counted_ptr) {
|
2022-02-28 04:57:36 +08:00
|
|
|
memnew_placement(r_dest, Ref<RefCounted>(p_ref_counted_ptr));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_ref_destroy(Ref<RefCounted> *p_instance) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_instance->~Ref();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_string_name_new_from_string(StringName *r_dest, const String *p_name) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, StringName(*p_name));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_new_from_string(NodePath *r_dest, const String *p_name) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, NodePath(*p_name));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_string_name_as_string(String *r_dest, const StringName *p_name) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, String(p_name->operator String()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_as_string(String *r_dest, const NodePath *p_np) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, String(p_np->operator String()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_byte_array_new_mem_copy(const uint8_t *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedByteArray);
|
|
|
|
PackedByteArray *array = reinterpret_cast<PackedByteArray *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
uint8_t *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(uint8_t));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_int32_array_new_mem_copy(const int32_t *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedInt32Array);
|
|
|
|
PackedInt32Array *array = reinterpret_cast<PackedInt32Array *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
int32_t *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(int32_t));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_int64_array_new_mem_copy(const int64_t *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedInt64Array);
|
|
|
|
PackedInt64Array *array = reinterpret_cast<PackedInt64Array *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
int64_t *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(int64_t));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_float32_array_new_mem_copy(const float *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedFloat32Array);
|
|
|
|
PackedFloat32Array *array = reinterpret_cast<PackedFloat32Array *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
float *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(float));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_float64_array_new_mem_copy(const double *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedFloat64Array);
|
|
|
|
PackedFloat64Array *array = reinterpret_cast<PackedFloat64Array *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
double *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(double));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_vector2_array_new_mem_copy(const Vector2 *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedVector2Array);
|
|
|
|
PackedVector2Array *array = reinterpret_cast<PackedVector2Array *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
Vector2 *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(Vector2));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_vector3_array_new_mem_copy(const Vector3 *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedVector3Array);
|
|
|
|
PackedVector3Array *array = reinterpret_cast<PackedVector3Array *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
Vector3 *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(Vector3));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_packed_color_array_new_mem_copy(const Color *p_src, int32_t p_length) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array ret;
|
|
|
|
memnew_placement(&ret, PackedColorArray);
|
|
|
|
PackedColorArray *array = reinterpret_cast<PackedColorArray *>(&ret);
|
|
|
|
array->resize(p_length);
|
|
|
|
Color *dst = array->ptrw();
|
|
|
|
memcpy(dst, p_src, p_length * sizeof(Color));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_string_array_add(PackedStringArray *r_dest, const String *p_element) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
r_dest->append(*p_element);
|
|
|
|
}
|
|
|
|
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
void godotsharp_callable_new_with_delegate(GCHandleIntPtr p_delegate_handle, void *p_trampoline,
|
|
|
|
const Object *p_object, Callable *r_callable) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
// TODO: Use pooling for ManagedCallable instances.
|
2022-10-01 00:22:11 +08:00
|
|
|
ObjectID objid = p_object ? p_object->get_instance_id() : ObjectID();
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
CallableCustom *managed_callable = memnew(ManagedCallable(p_delegate_handle, p_trampoline, objid));
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_callable, Callable(managed_callable));
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_callable_get_data_for_marshalling(const Callable *p_callable,
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
GCHandleIntPtr *r_delegate_handle, void **r_trampoline, Object **r_object, StringName *r_name) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
if (p_callable->is_custom()) {
|
|
|
|
CallableCustom *custom = p_callable->get_custom();
|
|
|
|
CallableCustom::CompareEqualFunc compare_equal_func = custom->get_compare_equal_func();
|
|
|
|
|
|
|
|
if (compare_equal_func == ManagedCallable::compare_equal_func_ptr) {
|
|
|
|
ManagedCallable *managed_callable = static_cast<ManagedCallable *>(custom);
|
|
|
|
*r_delegate_handle = managed_callable->get_delegate();
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
*r_trampoline = managed_callable->get_trampoline();
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
*r_object = nullptr;
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_name, StringName());
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return true;
|
|
|
|
} else if (compare_equal_func == SignalAwaiterCallable::compare_equal_func_ptr) {
|
|
|
|
SignalAwaiterCallable *signal_awaiter_callable = static_cast<SignalAwaiterCallable *>(custom);
|
2022-02-28 04:57:30 +08:00
|
|
|
*r_delegate_handle = { nullptr };
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
*r_trampoline = nullptr;
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
*r_object = ObjectDB::get_instance(signal_awaiter_callable->get_object());
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_name, StringName(signal_awaiter_callable->get_signal()));
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return true;
|
|
|
|
} else if (compare_equal_func == EventSignalCallable::compare_equal_func_ptr) {
|
|
|
|
EventSignalCallable *event_signal_callable = static_cast<EventSignalCallable *>(custom);
|
2022-02-28 04:57:30 +08:00
|
|
|
*r_delegate_handle = { nullptr };
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
*r_trampoline = nullptr;
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
*r_object = ObjectDB::get_instance(event_signal_callable->get_object());
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_name, StringName(event_signal_callable->get_signal()));
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Some other CallableCustom. We only support ManagedCallable.
|
2022-02-28 04:57:30 +08:00
|
|
|
*r_delegate_handle = { nullptr };
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
*r_trampoline = nullptr;
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
*r_object = nullptr;
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_name, StringName());
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return false;
|
|
|
|
} else {
|
2022-02-28 04:57:30 +08:00
|
|
|
*r_delegate_handle = { nullptr };
|
C#: Remove need for reflection to invoking callable delegates
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
2022-10-29 04:59:13 +08:00
|
|
|
*r_trampoline = nullptr;
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
*r_object = ObjectDB::get_instance(p_callable->get_object_id());
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_name, StringName(p_callable->get_method()));
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_variant godotsharp_callable_call(Callable *p_callable, const Variant **p_args, const int32_t p_arg_count, Callable::CallError *p_call_error) {
|
2021-12-29 06:25:16 +08:00
|
|
|
godot_variant ret;
|
|
|
|
memnew_placement(&ret, Variant);
|
|
|
|
|
|
|
|
Variant *ret_val = (Variant *)&ret;
|
|
|
|
|
|
|
|
p_callable->callp(p_args, p_arg_count, *ret_val, *p_call_error);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_callable_call_deferred(Callable *p_callable, const Variant **p_args, const int32_t p_arg_count) {
|
2021-12-29 06:25:16 +08:00
|
|
|
p_callable->call_deferredp(p_args, p_arg_count);
|
|
|
|
}
|
|
|
|
|
2022-12-06 11:58:16 +08:00
|
|
|
godot_color godotsharp_color_from_ok_hsl(float p_h, float p_s, float p_l, float p_alpha) {
|
|
|
|
godot_color ret;
|
|
|
|
Color *dest = (Color *)&ret;
|
|
|
|
memnew_placement(dest, Color(Color::from_ok_hsl(p_h, p_s, p_l, p_alpha)));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
// GDNative functions
|
|
|
|
|
|
|
|
// gdnative.h
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_method_bind_ptrcall(MethodBind *p_method_bind, Object *p_instance, const void **p_args, void *p_ret) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_method_bind->ptrcall(p_instance, p_args, p_ret);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_variant godotsharp_method_bind_call(MethodBind *p_method_bind, Object *p_instance, const godot_variant **p_args, const int32_t p_arg_count, Callable::CallError *p_call_error) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_variant ret;
|
|
|
|
memnew_placement(&ret, Variant());
|
|
|
|
|
|
|
|
Variant *ret_val = (Variant *)&ret;
|
|
|
|
|
|
|
|
*ret_val = p_method_bind->call(p_instance, (const Variant **)p_args, p_arg_count, *p_call_error);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
// variant.h
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_copy(godot_variant *r_dest, const Variant *p_src) {
|
2022-07-28 23:41:49 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_src));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_string_name(godot_variant *r_dest, const StringName *p_s) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_s));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_node_path(godot_variant *r_dest, const NodePath *p_np) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_np));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_object(godot_variant *r_dest, const Object *p_obj) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(p_obj));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_transform2d(godot_variant *r_dest, const Transform2D *p_t2d) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_t2d));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_basis(godot_variant *r_dest, const Basis *p_basis) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_basis));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_transform3d(godot_variant *r_dest, const Transform3D *p_trans) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_trans));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_projection(godot_variant *r_dest, const Projection *p_proj) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_proj));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_aabb(godot_variant *r_dest, const AABB *p_aabb) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_aabb));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_dictionary(godot_variant *r_dest, const Dictionary *p_dict) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_dict));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_array(godot_variant *r_dest, const Array *p_arr) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_arr));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_byte_array(godot_variant *r_dest, const PackedByteArray *p_pba) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pba));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_int32_array(godot_variant *r_dest, const PackedInt32Array *p_pia) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pia));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_int64_array(godot_variant *r_dest, const PackedInt64Array *p_pia) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pia));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_float32_array(godot_variant *r_dest, const PackedFloat32Array *p_pra) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pra));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_float64_array(godot_variant *r_dest, const PackedFloat64Array *p_pra) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pra));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_string_array(godot_variant *r_dest, const PackedStringArray *p_psa) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_psa));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_vector2_array(godot_variant *r_dest, const PackedVector2Array *p_pv2a) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pv2a));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_vector3_array(godot_variant *r_dest, const PackedVector3Array *p_pv3a) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pv3a));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_new_packed_color_array(godot_variant *r_dest, const PackedColorArray *p_pca) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Variant(*p_pca));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_variant_as_bool(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return p_self->operator bool();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int64_t godotsharp_variant_as_int(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return p_self->operator int64_t();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
double godotsharp_variant_as_float(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
return p_self->operator double();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_string godotsharp_variant_as_string(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_string raw_dest;
|
|
|
|
String *dest = (String *)&raw_dest;
|
|
|
|
memnew_placement(dest, String(p_self->operator String()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_vector2 godotsharp_variant_as_vector2(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_vector2 raw_dest;
|
|
|
|
Vector2 *dest = (Vector2 *)&raw_dest;
|
|
|
|
memnew_placement(dest, Vector2(p_self->operator Vector2()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_vector2i godotsharp_variant_as_vector2i(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_vector2i raw_dest;
|
|
|
|
Vector2i *dest = (Vector2i *)&raw_dest;
|
|
|
|
memnew_placement(dest, Vector2i(p_self->operator Vector2i()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_rect2 godotsharp_variant_as_rect2(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_rect2 raw_dest;
|
|
|
|
Rect2 *dest = (Rect2 *)&raw_dest;
|
|
|
|
memnew_placement(dest, Rect2(p_self->operator Rect2()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_rect2i godotsharp_variant_as_rect2i(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_rect2i raw_dest;
|
|
|
|
Rect2i *dest = (Rect2i *)&raw_dest;
|
|
|
|
memnew_placement(dest, Rect2i(p_self->operator Rect2i()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_vector3 godotsharp_variant_as_vector3(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_vector3 raw_dest;
|
|
|
|
Vector3 *dest = (Vector3 *)&raw_dest;
|
|
|
|
memnew_placement(dest, Vector3(p_self->operator Vector3()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_vector3i godotsharp_variant_as_vector3i(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_vector3i raw_dest;
|
|
|
|
Vector3i *dest = (Vector3i *)&raw_dest;
|
|
|
|
memnew_placement(dest, Vector3i(p_self->operator Vector3i()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_transform2d godotsharp_variant_as_transform2d(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_transform2d raw_dest;
|
|
|
|
Transform2D *dest = (Transform2D *)&raw_dest;
|
|
|
|
memnew_placement(dest, Transform2D(p_self->operator Transform2D()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_vector4 godotsharp_variant_as_vector4(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_vector4 raw_dest;
|
|
|
|
Vector4 *dest = (Vector4 *)&raw_dest;
|
|
|
|
memnew_placement(dest, Vector4(p_self->operator Vector4()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_vector4i godotsharp_variant_as_vector4i(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_vector4i raw_dest;
|
|
|
|
Vector4i *dest = (Vector4i *)&raw_dest;
|
|
|
|
memnew_placement(dest, Vector4i(p_self->operator Vector4i()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_plane godotsharp_variant_as_plane(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_plane raw_dest;
|
|
|
|
Plane *dest = (Plane *)&raw_dest;
|
|
|
|
memnew_placement(dest, Plane(p_self->operator Plane()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_quaternion godotsharp_variant_as_quaternion(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_quaternion raw_dest;
|
|
|
|
Quaternion *dest = (Quaternion *)&raw_dest;
|
|
|
|
memnew_placement(dest, Quaternion(p_self->operator Quaternion()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_aabb godotsharp_variant_as_aabb(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_aabb raw_dest;
|
|
|
|
AABB *dest = (AABB *)&raw_dest;
|
|
|
|
memnew_placement(dest, AABB(p_self->operator ::AABB()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_basis godotsharp_variant_as_basis(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_basis raw_dest;
|
|
|
|
Basis *dest = (Basis *)&raw_dest;
|
|
|
|
memnew_placement(dest, Basis(p_self->operator Basis()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_transform3d godotsharp_variant_as_transform3d(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_transform3d raw_dest;
|
|
|
|
Transform3D *dest = (Transform3D *)&raw_dest;
|
|
|
|
memnew_placement(dest, Transform3D(p_self->operator Transform3D()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_projection godotsharp_variant_as_projection(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_projection raw_dest;
|
|
|
|
Projection *dest = (Projection *)&raw_dest;
|
|
|
|
memnew_placement(dest, Projection(p_self->operator Projection()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_color godotsharp_variant_as_color(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_color raw_dest;
|
|
|
|
Color *dest = (Color *)&raw_dest;
|
|
|
|
memnew_placement(dest, Color(p_self->operator Color()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_string_name godotsharp_variant_as_string_name(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_string_name raw_dest;
|
|
|
|
StringName *dest = (StringName *)&raw_dest;
|
|
|
|
memnew_placement(dest, StringName(p_self->operator StringName()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_node_path godotsharp_variant_as_node_path(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_node_path raw_dest;
|
|
|
|
NodePath *dest = (NodePath *)&raw_dest;
|
|
|
|
memnew_placement(dest, NodePath(p_self->operator NodePath()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_rid godotsharp_variant_as_rid(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_rid raw_dest;
|
|
|
|
RID *dest = (RID *)&raw_dest;
|
|
|
|
memnew_placement(dest, RID(p_self->operator ::RID()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_callable godotsharp_variant_as_callable(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_callable raw_dest;
|
|
|
|
Callable *dest = (Callable *)&raw_dest;
|
|
|
|
memnew_placement(dest, Callable(p_self->operator Callable()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_signal godotsharp_variant_as_signal(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_signal raw_dest;
|
|
|
|
Signal *dest = (Signal *)&raw_dest;
|
|
|
|
memnew_placement(dest, Signal(p_self->operator Signal()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_dictionary godotsharp_variant_as_dictionary(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_dictionary raw_dest;
|
|
|
|
Dictionary *dest = (Dictionary *)&raw_dest;
|
|
|
|
memnew_placement(dest, Dictionary(p_self->operator Dictionary()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_array godotsharp_variant_as_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_array raw_dest;
|
|
|
|
Array *dest = (Array *)&raw_dest;
|
|
|
|
memnew_placement(dest, Array(p_self->operator Array()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_byte_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedByteArray *dest = (PackedByteArray *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedByteArray(p_self->operator PackedByteArray()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_int32_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedInt32Array *dest = (PackedInt32Array *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedInt32Array(p_self->operator PackedInt32Array()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_int64_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedInt64Array *dest = (PackedInt64Array *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedInt64Array(p_self->operator PackedInt64Array()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_float32_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedFloat32Array *dest = (PackedFloat32Array *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedFloat32Array(p_self->operator PackedFloat32Array()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_float64_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedFloat64Array *dest = (PackedFloat64Array *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedFloat64Array(p_self->operator PackedFloat64Array()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_string_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedStringArray *dest = (PackedStringArray *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedStringArray(p_self->operator PackedStringArray()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_vector2_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedVector2Array *dest = (PackedVector2Array *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedVector2Array(p_self->operator PackedVector2Array()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_vector3_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedVector3Array *dest = (PackedVector3Array *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedVector3Array(p_self->operator PackedVector3Array()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_packed_array godotsharp_variant_as_packed_color_array(const Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
godot_packed_array raw_dest;
|
|
|
|
PackedColorArray *dest = (PackedColorArray *)&raw_dest;
|
|
|
|
memnew_placement(dest, PackedColorArray(p_self->operator PackedColorArray()));
|
|
|
|
return raw_dest;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_variant_equals(const godot_variant *p_a, const godot_variant *p_b) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return *reinterpret_cast<const Variant *>(p_a) == *reinterpret_cast<const Variant *>(p_b);
|
|
|
|
}
|
|
|
|
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
// string.h
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_string_new_with_utf16_chars(String *r_dest, const char16_t *p_contents) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, String());
|
|
|
|
r_dest->parse_utf16(p_contents);
|
|
|
|
}
|
|
|
|
|
|
|
|
// string_name.h
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_string_name_new_copy(StringName *r_dest, const StringName *p_src) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, StringName(*p_src));
|
|
|
|
}
|
|
|
|
|
|
|
|
// node_path.h
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_new_copy(NodePath *r_dest, const NodePath *p_src) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, NodePath(*p_src));
|
|
|
|
}
|
|
|
|
|
|
|
|
// array.h
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_new(Array *r_dest) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_dest, Array);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_new_copy(Array *r_dest, const Array *p_src) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Array(*p_src));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
godot_variant *godotsharp_array_ptrw(godot_array *p_self) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return reinterpret_cast<godot_variant *>(&reinterpret_cast<Array *>(p_self)->operator[](0));
|
|
|
|
}
|
|
|
|
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
// dictionary.h
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_new(Dictionary *r_dest) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_dest, Dictionary);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_new_copy(Dictionary *r_dest, const Dictionary *p_src) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
memnew_placement(r_dest, Dictionary(*p_src));
|
|
|
|
}
|
|
|
|
|
|
|
|
// destroy functions
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_byte_array_destroy(PackedByteArray *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedByteArray();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_int32_array_destroy(PackedInt32Array *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedInt32Array();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_int64_array_destroy(PackedInt64Array *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedInt64Array();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_float32_array_destroy(PackedFloat32Array *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedFloat32Array();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_float64_array_destroy(PackedFloat64Array *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedFloat64Array();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_string_array_destroy(PackedStringArray *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedStringArray();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_vector2_array_destroy(PackedVector2Array *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedVector2Array();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_vector3_array_destroy(PackedVector3Array *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedVector3Array();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_packed_color_array_destroy(PackedColorArray *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~PackedColorArray();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_variant_destroy(Variant *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~Variant();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_string_destroy(String *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~String();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_string_name_destroy(StringName *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~StringName();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_destroy(NodePath *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~NodePath();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_signal_destroy(Signal *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~Signal();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_callable_destroy(Callable *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~Callable();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_destroy(Array *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~Array();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_destroy(Dictionary *p_self) {
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
p_self->~Dictionary();
|
|
|
|
}
|
|
|
|
|
2021-09-13 01:49:39 +08:00
|
|
|
// Array
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_array_add(Array *p_self, const Variant *p_item) {
|
2021-09-13 01:49:39 +08:00
|
|
|
p_self->append(*p_item);
|
|
|
|
return p_self->size();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_duplicate(const Array *p_self, bool p_deep, Array *r_dest) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_dest, Array(p_self->duplicate(p_deep)));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_array_index_of(const Array *p_self, const Variant *p_item) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return p_self->find(*p_item);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_insert(Array *p_self, int32_t p_index, const Variant *p_item) {
|
2021-09-13 01:49:39 +08:00
|
|
|
p_self->insert(p_index, *p_item);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_remove_at(Array *p_self, int32_t p_index) {
|
2021-09-13 01:49:39 +08:00
|
|
|
p_self->remove_at(p_index);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_array_resize(Array *p_self, int32_t p_new_size) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return (int32_t)p_self->resize(p_new_size);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_shuffle(Array *p_self) {
|
2021-09-13 01:49:39 +08:00
|
|
|
p_self->shuffle();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_array_to_string(const Array *p_self, String *r_str) {
|
|
|
|
*r_str = Variant(*p_self).operator String();
|
|
|
|
}
|
|
|
|
|
2021-09-13 01:49:39 +08:00
|
|
|
// Dictionary
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_dictionary_try_get_value(const Dictionary *p_self, const Variant *p_key, Variant *r_value) {
|
2021-09-13 01:49:39 +08:00
|
|
|
const Variant *ret = p_self->getptr(*p_key);
|
|
|
|
if (ret == nullptr) {
|
|
|
|
memnew_placement(r_value, Variant());
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
memnew_placement(r_value, Variant(*ret));
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_set_value(Dictionary *p_self, const Variant *p_key, const Variant *p_value) {
|
2021-09-13 01:49:39 +08:00
|
|
|
p_self->operator[](*p_key) = *p_value;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_keys(const Dictionary *p_self, Array *r_dest) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_dest, Array(p_self->keys()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_values(const Dictionary *p_self, Array *r_dest) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_dest, Array(p_self->values()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_dictionary_count(const Dictionary *p_self) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return p_self->size();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_key_value_pair_at(const Dictionary *p_self, int32_t p_index, Variant *r_key, Variant *r_value) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_key, Variant(p_self->get_key_at_index(p_index)));
|
|
|
|
memnew_placement(r_value, Variant(p_self->get_value_at_index(p_index)));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_add(Dictionary *p_self, const Variant *p_key, const Variant *p_value) {
|
2021-09-13 01:49:39 +08:00
|
|
|
p_self->operator[](*p_key) = *p_value;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_clear(Dictionary *p_self) {
|
2021-09-13 01:49:39 +08:00
|
|
|
p_self->clear();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_dictionary_contains_key(const Dictionary *p_self, const Variant *p_key) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return p_self->has(*p_key);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_duplicate(const Dictionary *p_self, bool p_deep, Dictionary *r_dest) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_dest, Dictionary(p_self->duplicate(p_deep)));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_dictionary_remove_key(Dictionary *p_self, const Variant *p_key) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return p_self->erase(*p_key);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_dictionary_to_string(const Dictionary *p_self, String *r_str) {
|
|
|
|
*r_str = Variant(*p_self).operator String();
|
|
|
|
}
|
|
|
|
|
|
|
|
void godotsharp_string_simplify_path(const String *p_self, String *r_simplified_path) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_simplified_path, String(p_self->simplify_path()));
|
|
|
|
}
|
|
|
|
|
2022-08-30 17:36:24 +08:00
|
|
|
void godotsharp_string_to_camel_case(const String *p_self, String *r_camel_case) {
|
|
|
|
memnew_placement(r_camel_case, String(p_self->to_camel_case()));
|
|
|
|
}
|
|
|
|
|
|
|
|
void godotsharp_string_to_pascal_case(const String *p_self, String *r_pascal_case) {
|
|
|
|
memnew_placement(r_pascal_case, String(p_self->to_pascal_case()));
|
|
|
|
}
|
|
|
|
|
|
|
|
void godotsharp_string_to_snake_case(const String *p_self, String *r_snake_case) {
|
|
|
|
memnew_placement(r_snake_case, String(p_self->to_snake_case()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_get_as_property_path(const NodePath *p_ptr, NodePath *r_dest) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_dest, NodePath(p_ptr->get_as_property_path()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_get_concatenated_names(const NodePath *p_self, String *r_subnames) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_subnames, String(p_self->get_concatenated_names()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_get_concatenated_subnames(const NodePath *p_self, String *r_subnames) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_subnames, String(p_self->get_concatenated_subnames()));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_get_name(const NodePath *p_self, uint32_t p_idx, String *r_name) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_name, String(p_self->get_name(p_idx)));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_node_path_get_name_count(const NodePath *p_self) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return p_self->get_name_count();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_node_path_get_subname(const NodePath *p_self, uint32_t p_idx, String *r_subname) {
|
2021-09-13 01:49:39 +08:00
|
|
|
memnew_placement(r_subname, String(p_self->get_subname(p_idx)));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_node_path_get_subname_count(const NodePath *p_self) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return p_self->get_subname_count();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
bool godotsharp_node_path_is_absolute(const NodePath *p_self) {
|
2021-09-13 01:49:39 +08:00
|
|
|
return p_self->is_absolute();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_randomize() {
|
2021-09-13 01:50:13 +08:00
|
|
|
Math::randomize();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
uint32_t godotsharp_randi() {
|
2021-09-13 01:50:13 +08:00
|
|
|
return Math::rand();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
float godotsharp_randf() {
|
2021-09-13 01:50:13 +08:00
|
|
|
return Math::randf();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int32_t godotsharp_randi_range(int32_t p_from, int32_t p_to) {
|
2021-09-13 01:50:13 +08:00
|
|
|
return Math::random(p_from, p_to);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
double godotsharp_randf_range(double p_from, double p_to) {
|
2021-09-13 01:50:13 +08:00
|
|
|
return Math::random(p_from, p_to);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
double godotsharp_randfn(double p_mean, double p_deviation) {
|
2021-09-13 01:50:13 +08:00
|
|
|
return Math::randfn(p_mean, p_deviation);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_seed(uint64_t p_seed) {
|
2021-09-13 01:50:13 +08:00
|
|
|
Math::seed(p_seed);
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
uint32_t godotsharp_rand_from_seed(uint64_t p_seed, uint64_t *r_new_seed) {
|
2021-09-13 01:50:13 +08:00
|
|
|
uint32_t ret = Math::rand_from_seed(&p_seed);
|
|
|
|
*r_new_seed = p_seed;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_weakref(Object *p_ptr, Ref<RefCounted> *r_weak_ref) {
|
2021-09-13 01:50:13 +08:00
|
|
|
if (!p_ptr) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
Ref<WeakRef> wref;
|
|
|
|
RefCounted *rc = Object::cast_to<RefCounted>(p_ptr);
|
|
|
|
|
|
|
|
if (rc) {
|
|
|
|
Ref<RefCounted> r = rc;
|
|
|
|
if (!r.is_valid()) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
wref.instantiate();
|
|
|
|
wref->set_ref(r);
|
|
|
|
} else {
|
|
|
|
wref.instantiate();
|
|
|
|
wref->set_obj(p_ptr);
|
|
|
|
}
|
|
|
|
|
|
|
|
memnew_placement(r_weak_ref, Ref<RefCounted>(wref));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_str(const godot_array *p_what, godot_string *r_ret) {
|
2021-09-13 01:50:13 +08:00
|
|
|
String &str = *memnew_placement(r_ret, String);
|
|
|
|
const Array &what = *reinterpret_cast<const Array *>(p_what);
|
|
|
|
|
|
|
|
for (int i = 0; i < what.size(); i++) {
|
|
|
|
String os = what[i].operator String();
|
|
|
|
|
|
|
|
if (i == 0) {
|
|
|
|
str = os;
|
|
|
|
} else {
|
|
|
|
str += os;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_print(const godot_string *p_what) {
|
2021-09-13 01:50:13 +08:00
|
|
|
print_line(*reinterpret_cast<const String *>(p_what));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_print_rich(const godot_string *p_what) {
|
2021-09-13 01:50:13 +08:00
|
|
|
print_line_rich(*reinterpret_cast<const String *>(p_what));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_printerr(const godot_string *p_what) {
|
2021-09-13 01:50:13 +08:00
|
|
|
print_error(*reinterpret_cast<const String *>(p_what));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_printt(const godot_string *p_what) {
|
2021-09-13 01:50:13 +08:00
|
|
|
print_line(*reinterpret_cast<const String *>(p_what));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_prints(const godot_string *p_what) {
|
2021-09-13 01:50:13 +08:00
|
|
|
print_line(*reinterpret_cast<const String *>(p_what));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_printraw(const godot_string *p_what) {
|
2021-09-13 01:50:13 +08:00
|
|
|
OS::get_singleton()->print("%s", reinterpret_cast<const String *>(p_what)->utf8().get_data());
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_pusherror(const godot_string *p_str) {
|
2021-09-13 01:50:13 +08:00
|
|
|
ERR_PRINT(*reinterpret_cast<const String *>(p_str));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_pushwarning(const godot_string *p_str) {
|
2021-09-13 01:50:13 +08:00
|
|
|
WARN_PRINT(*reinterpret_cast<const String *>(p_str));
|
|
|
|
}
|
|
|
|
|
2022-08-13 23:45:42 +08:00
|
|
|
void godotsharp_var_to_str(const godot_variant *p_var, godot_string *r_ret) {
|
2021-09-13 01:50:13 +08:00
|
|
|
const Variant &var = *reinterpret_cast<const Variant *>(p_var);
|
|
|
|
String &vars = *memnew_placement(r_ret, String);
|
|
|
|
VariantWriter::write_to_string(var, vars);
|
|
|
|
}
|
|
|
|
|
2022-08-13 23:45:42 +08:00
|
|
|
void godotsharp_str_to_var(const godot_string *p_str, godot_variant *r_ret) {
|
2021-09-13 01:50:13 +08:00
|
|
|
Variant ret;
|
|
|
|
|
|
|
|
VariantParser::StreamString ss;
|
|
|
|
ss.s = *reinterpret_cast<const String *>(p_str);
|
|
|
|
|
|
|
|
String errs;
|
|
|
|
int line;
|
|
|
|
Error err = VariantParser::parse(&ss, ret, errs, line);
|
|
|
|
if (err != OK) {
|
|
|
|
String err_str = "Parse error at line " + itos(line) + ": " + errs + ".";
|
|
|
|
ERR_PRINT(err_str);
|
|
|
|
ret = err_str;
|
|
|
|
}
|
|
|
|
memnew_placement(r_ret, Variant(ret));
|
|
|
|
}
|
|
|
|
|
2022-08-13 23:45:42 +08:00
|
|
|
void godotsharp_var_to_bytes(const godot_variant *p_var, bool p_full_objects, godot_packed_array *r_bytes) {
|
2021-09-13 01:50:13 +08:00
|
|
|
const Variant &var = *reinterpret_cast<const Variant *>(p_var);
|
|
|
|
PackedByteArray &bytes = *memnew_placement(r_bytes, PackedByteArray);
|
|
|
|
|
|
|
|
int len;
|
|
|
|
Error err = encode_variant(var, nullptr, len, p_full_objects);
|
|
|
|
ERR_FAIL_COND_MSG(err != OK, "Unexpected error encoding variable to bytes, likely unserializable type found (Object or RID).");
|
|
|
|
|
|
|
|
bytes.resize(len);
|
|
|
|
encode_variant(var, bytes.ptrw(), len, p_full_objects);
|
|
|
|
}
|
|
|
|
|
2022-08-13 23:45:42 +08:00
|
|
|
void godotsharp_bytes_to_var(const godot_packed_array *p_bytes, bool p_allow_objects, godot_variant *r_ret) {
|
2021-09-13 01:50:13 +08:00
|
|
|
const PackedByteArray *bytes = reinterpret_cast<const PackedByteArray *>(p_bytes);
|
|
|
|
Variant ret;
|
|
|
|
Error err = decode_variant(ret, bytes->ptr(), bytes->size(), nullptr, p_allow_objects);
|
|
|
|
if (err != OK) {
|
|
|
|
ret = RTR("Not enough bytes for decoding bytes, or invalid format.");
|
|
|
|
}
|
|
|
|
memnew_placement(r_ret, Variant(ret));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
int godotsharp_hash(const godot_variant *p_var) {
|
2021-09-13 01:50:13 +08:00
|
|
|
return reinterpret_cast<const Variant *>(p_var)->hash();
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_convert(const godot_variant *p_what, int32_t p_type, godot_variant *r_ret) {
|
2021-09-13 01:50:13 +08:00
|
|
|
const Variant *args[1] = { reinterpret_cast<const Variant *>(p_what) };
|
|
|
|
Callable::CallError ce;
|
|
|
|
Variant ret;
|
|
|
|
Variant::construct(Variant::Type(p_type), ret, args, 1, ce);
|
|
|
|
if (ce.error != Callable::CallError::CALL_OK) {
|
|
|
|
memnew_placement(r_ret, Variant);
|
|
|
|
ERR_FAIL_MSG("Unable to convert parameter from '" +
|
2021-09-13 02:23:05 +08:00
|
|
|
Variant::get_type_name(reinterpret_cast<const Variant *>(p_what)->get_type()) +
|
|
|
|
"' to '" + Variant::get_type_name(Variant::Type(p_type)) + "'.");
|
2021-09-13 01:50:13 +08:00
|
|
|
}
|
|
|
|
memnew_placement(r_ret, Variant(ret));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
Object *godotsharp_instance_from_id(uint64_t p_instance_id) {
|
2021-09-13 01:50:13 +08:00
|
|
|
return ObjectDB::get_instance(ObjectID(p_instance_id));
|
|
|
|
}
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
void godotsharp_object_to_string(Object *p_ptr, godot_string *r_str) {
|
2021-09-13 01:50:13 +08:00
|
|
|
#ifdef DEBUG_ENABLED
|
|
|
|
// Cannot happen in C#; would get an ObjectDisposedException instead.
|
|
|
|
CRASH_COND(p_ptr == nullptr);
|
|
|
|
#endif
|
|
|
|
// Can't call 'Object::to_string()' here, as that can end up calling 'ToString' again resulting in an endless circular loop.
|
|
|
|
memnew_placement(r_str,
|
2022-07-25 06:15:20 +08:00
|
|
|
String("<" + p_ptr->get_class() + "#" + itos(p_ptr->get_instance_id()) + ">"));
|
2021-09-13 01:50:13 +08:00
|
|
|
}
|
|
|
|
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
#ifdef __cplusplus
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2022-08-05 09:32:59 +08:00
|
|
|
// The order in this array must match the declaration order of
|
|
|
|
// the methods in 'GodotSharp/Core/NativeInterop/NativeFuncs.cs'.
|
|
|
|
static const void *unmanaged_callbacks[]{
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
(void *)godotsharp_method_bind_get_method,
|
|
|
|
(void *)godotsharp_get_class_constructor,
|
|
|
|
(void *)godotsharp_engine_get_singleton,
|
2022-02-28 04:57:39 +08:00
|
|
|
(void *)godotsharp_stack_info_vector_resize,
|
|
|
|
(void *)godotsharp_stack_info_vector_destroy,
|
|
|
|
(void *)godotsharp_internal_script_debugger_send_error,
|
|
|
|
(void *)godotsharp_internal_script_debugger_is_active,
|
2022-02-28 04:57:30 +08:00
|
|
|
(void *)godotsharp_internal_object_get_associated_gchandle,
|
2021-09-13 02:23:05 +08:00
|
|
|
(void *)godotsharp_internal_object_disposed,
|
|
|
|
(void *)godotsharp_internal_refcounted_disposed,
|
|
|
|
(void *)godotsharp_internal_signal_awaiter_connect,
|
|
|
|
(void *)godotsharp_internal_tie_native_managed_to_unmanaged,
|
|
|
|
(void *)godotsharp_internal_tie_user_managed_to_unmanaged,
|
|
|
|
(void *)godotsharp_internal_tie_managed_to_unmanaged_with_pre_setup,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_internal_unmanaged_get_script_instance_managed,
|
|
|
|
(void *)godotsharp_internal_unmanaged_get_instance_binding_managed,
|
|
|
|
(void *)godotsharp_internal_unmanaged_instance_binding_create_managed,
|
2021-09-13 02:23:05 +08:00
|
|
|
(void *)godotsharp_internal_new_csharp_script,
|
2022-05-28 10:56:46 +08:00
|
|
|
(void *)godotsharp_internal_script_load,
|
2022-02-28 04:57:36 +08:00
|
|
|
(void *)godotsharp_internal_reload_registered_script,
|
2021-09-13 02:23:05 +08:00
|
|
|
(void *)godotsharp_array_filter_godot_objects_by_native,
|
|
|
|
(void *)godotsharp_array_filter_godot_objects_by_non_native,
|
2022-02-28 04:57:36 +08:00
|
|
|
(void *)godotsharp_ref_new_from_ref_counted_ptr,
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
(void *)godotsharp_ref_destroy,
|
|
|
|
(void *)godotsharp_string_name_new_from_string,
|
|
|
|
(void *)godotsharp_node_path_new_from_string,
|
|
|
|
(void *)godotsharp_string_name_as_string,
|
|
|
|
(void *)godotsharp_node_path_as_string,
|
|
|
|
(void *)godotsharp_packed_byte_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_int32_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_int64_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_float32_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_float64_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_vector2_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_vector3_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_color_array_new_mem_copy,
|
|
|
|
(void *)godotsharp_packed_string_array_add,
|
|
|
|
(void *)godotsharp_callable_new_with_delegate,
|
|
|
|
(void *)godotsharp_callable_get_data_for_marshalling,
|
2021-12-29 06:25:16 +08:00
|
|
|
(void *)godotsharp_callable_call,
|
|
|
|
(void *)godotsharp_callable_call_deferred,
|
2022-12-06 11:58:16 +08:00
|
|
|
(void *)godotsharp_color_from_ok_hsl,
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
(void *)godotsharp_method_bind_ptrcall,
|
|
|
|
(void *)godotsharp_method_bind_call,
|
|
|
|
(void *)godotsharp_variant_new_string_name,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_variant_new_copy,
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
(void *)godotsharp_variant_new_node_path,
|
|
|
|
(void *)godotsharp_variant_new_object,
|
|
|
|
(void *)godotsharp_variant_new_transform2d,
|
|
|
|
(void *)godotsharp_variant_new_basis,
|
|
|
|
(void *)godotsharp_variant_new_transform3d,
|
|
|
|
(void *)godotsharp_variant_new_projection,
|
|
|
|
(void *)godotsharp_variant_new_aabb,
|
|
|
|
(void *)godotsharp_variant_new_dictionary,
|
|
|
|
(void *)godotsharp_variant_new_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_byte_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_int32_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_int64_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_float32_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_float64_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_string_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_vector2_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_vector3_array,
|
|
|
|
(void *)godotsharp_variant_new_packed_color_array,
|
|
|
|
(void *)godotsharp_variant_as_bool,
|
|
|
|
(void *)godotsharp_variant_as_int,
|
|
|
|
(void *)godotsharp_variant_as_float,
|
|
|
|
(void *)godotsharp_variant_as_string,
|
|
|
|
(void *)godotsharp_variant_as_vector2,
|
|
|
|
(void *)godotsharp_variant_as_vector2i,
|
|
|
|
(void *)godotsharp_variant_as_rect2,
|
|
|
|
(void *)godotsharp_variant_as_rect2i,
|
|
|
|
(void *)godotsharp_variant_as_vector3,
|
|
|
|
(void *)godotsharp_variant_as_vector3i,
|
|
|
|
(void *)godotsharp_variant_as_transform2d,
|
|
|
|
(void *)godotsharp_variant_as_vector4,
|
|
|
|
(void *)godotsharp_variant_as_vector4i,
|
|
|
|
(void *)godotsharp_variant_as_plane,
|
|
|
|
(void *)godotsharp_variant_as_quaternion,
|
|
|
|
(void *)godotsharp_variant_as_aabb,
|
|
|
|
(void *)godotsharp_variant_as_basis,
|
|
|
|
(void *)godotsharp_variant_as_transform3d,
|
|
|
|
(void *)godotsharp_variant_as_projection,
|
|
|
|
(void *)godotsharp_variant_as_color,
|
|
|
|
(void *)godotsharp_variant_as_string_name,
|
|
|
|
(void *)godotsharp_variant_as_node_path,
|
|
|
|
(void *)godotsharp_variant_as_rid,
|
|
|
|
(void *)godotsharp_variant_as_callable,
|
|
|
|
(void *)godotsharp_variant_as_signal,
|
|
|
|
(void *)godotsharp_variant_as_dictionary,
|
|
|
|
(void *)godotsharp_variant_as_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_byte_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_int32_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_int64_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_float32_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_float64_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_string_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_vector2_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_vector3_array,
|
|
|
|
(void *)godotsharp_variant_as_packed_color_array,
|
2021-09-13 01:49:39 +08:00
|
|
|
(void *)godotsharp_variant_equals,
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
(void *)godotsharp_string_new_with_utf16_chars,
|
|
|
|
(void *)godotsharp_string_name_new_copy,
|
|
|
|
(void *)godotsharp_node_path_new_copy,
|
2021-09-13 01:49:39 +08:00
|
|
|
(void *)godotsharp_array_new,
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
(void *)godotsharp_array_new_copy,
|
2021-09-13 01:49:39 +08:00
|
|
|
(void *)godotsharp_array_ptrw,
|
|
|
|
(void *)godotsharp_dictionary_new,
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
(void *)godotsharp_dictionary_new_copy,
|
|
|
|
(void *)godotsharp_packed_byte_array_destroy,
|
|
|
|
(void *)godotsharp_packed_int32_array_destroy,
|
|
|
|
(void *)godotsharp_packed_int64_array_destroy,
|
|
|
|
(void *)godotsharp_packed_float32_array_destroy,
|
|
|
|
(void *)godotsharp_packed_float64_array_destroy,
|
|
|
|
(void *)godotsharp_packed_string_array_destroy,
|
|
|
|
(void *)godotsharp_packed_vector2_array_destroy,
|
|
|
|
(void *)godotsharp_packed_vector3_array_destroy,
|
|
|
|
(void *)godotsharp_packed_color_array_destroy,
|
|
|
|
(void *)godotsharp_variant_destroy,
|
|
|
|
(void *)godotsharp_string_destroy,
|
|
|
|
(void *)godotsharp_string_name_destroy,
|
|
|
|
(void *)godotsharp_node_path_destroy,
|
|
|
|
(void *)godotsharp_signal_destroy,
|
|
|
|
(void *)godotsharp_callable_destroy,
|
|
|
|
(void *)godotsharp_array_destroy,
|
2021-09-13 01:49:39 +08:00
|
|
|
(void *)godotsharp_dictionary_destroy,
|
|
|
|
(void *)godotsharp_array_add,
|
|
|
|
(void *)godotsharp_array_duplicate,
|
|
|
|
(void *)godotsharp_array_index_of,
|
|
|
|
(void *)godotsharp_array_insert,
|
|
|
|
(void *)godotsharp_array_remove_at,
|
|
|
|
(void *)godotsharp_array_resize,
|
|
|
|
(void *)godotsharp_array_shuffle,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_array_to_string,
|
2021-09-13 01:49:39 +08:00
|
|
|
(void *)godotsharp_dictionary_try_get_value,
|
|
|
|
(void *)godotsharp_dictionary_set_value,
|
|
|
|
(void *)godotsharp_dictionary_keys,
|
|
|
|
(void *)godotsharp_dictionary_values,
|
|
|
|
(void *)godotsharp_dictionary_count,
|
|
|
|
(void *)godotsharp_dictionary_key_value_pair_at,
|
|
|
|
(void *)godotsharp_dictionary_add,
|
|
|
|
(void *)godotsharp_dictionary_clear,
|
|
|
|
(void *)godotsharp_dictionary_contains_key,
|
|
|
|
(void *)godotsharp_dictionary_duplicate,
|
|
|
|
(void *)godotsharp_dictionary_remove_key,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_dictionary_to_string,
|
2021-09-13 01:49:39 +08:00
|
|
|
(void *)godotsharp_string_simplify_path,
|
2022-08-30 17:36:24 +08:00
|
|
|
(void *)godotsharp_string_to_camel_case,
|
|
|
|
(void *)godotsharp_string_to_pascal_case,
|
|
|
|
(void *)godotsharp_string_to_snake_case,
|
2021-09-13 01:49:39 +08:00
|
|
|
(void *)godotsharp_node_path_get_as_property_path,
|
|
|
|
(void *)godotsharp_node_path_get_concatenated_names,
|
|
|
|
(void *)godotsharp_node_path_get_concatenated_subnames,
|
|
|
|
(void *)godotsharp_node_path_get_name,
|
|
|
|
(void *)godotsharp_node_path_get_name_count,
|
|
|
|
(void *)godotsharp_node_path_get_subname,
|
|
|
|
(void *)godotsharp_node_path_get_subname_count,
|
2021-09-13 01:50:13 +08:00
|
|
|
(void *)godotsharp_node_path_is_absolute,
|
2022-08-13 23:45:42 +08:00
|
|
|
(void *)godotsharp_bytes_to_var,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_convert,
|
|
|
|
(void *)godotsharp_hash,
|
|
|
|
(void *)godotsharp_instance_from_id,
|
|
|
|
(void *)godotsharp_print,
|
|
|
|
(void *)godotsharp_print_rich,
|
|
|
|
(void *)godotsharp_printerr,
|
|
|
|
(void *)godotsharp_printraw,
|
|
|
|
(void *)godotsharp_prints,
|
|
|
|
(void *)godotsharp_printt,
|
2021-09-13 01:50:13 +08:00
|
|
|
(void *)godotsharp_randf,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_randi,
|
|
|
|
(void *)godotsharp_randomize,
|
2021-09-13 01:50:13 +08:00
|
|
|
(void *)godotsharp_randf_range,
|
|
|
|
(void *)godotsharp_randfn,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_randi_range,
|
2021-09-13 01:50:13 +08:00
|
|
|
(void *)godotsharp_rand_from_seed,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_seed,
|
2021-09-13 01:50:13 +08:00
|
|
|
(void *)godotsharp_weakref,
|
|
|
|
(void *)godotsharp_str,
|
2022-08-13 23:45:42 +08:00
|
|
|
(void *)godotsharp_str_to_var,
|
|
|
|
(void *)godotsharp_var_to_bytes,
|
|
|
|
(void *)godotsharp_var_to_str,
|
2022-08-05 09:32:59 +08:00
|
|
|
(void *)godotsharp_pusherror,
|
|
|
|
(void *)godotsharp_pushwarning,
|
2021-09-13 01:50:13 +08:00
|
|
|
(void *)godotsharp_object_to_string,
|
C#: Move marshaling logic and generated glue to C#
We will be progressively moving most code to C#.
The plan is to only use Mono's embedding APIs to set things at launch.
This will make it much easier to later support CoreCLR too which
doesn't have rich embedding APIs.
Additionally the code in C# is more maintainable and makes it easier
to implement new features, e.g.: runtime codegen which we could use to
avoid using reflection for marshaling everytime a field, property or
method is accessed.
SOME NOTES ON INTEROP
We make the same assumptions as GDNative about the size of the Godot
structures we use. We take it a bit further by also assuming the layout
of fields in some cases, which is riskier but let's us squeeze out some
performance by avoiding unnecessary managed to native calls.
Code that deals with native structs is less safe than before as there's
no RAII and copy constructors in C#. It's like using the GDNative C API
directly. One has to take special care to free values they own.
Perhaps we could use roslyn analyzers to check this, but I don't know
any that uses attributes to determine what's owned or borrowed.
As to why we maily use pointers for native structs instead of ref/out:
- AFAIK (and confirmed with a benchmark) ref/out are pinned
during P/Invoke calls and that has a cost.
- Native struct fields can't be ref/out in the first place.
- A `using` local can't be passed as ref/out, only `in`. Calling a
method or property on an `in` value makes a silent copy, so we want
to avoid `in`.
REGARDING THE BUILD SYSTEM
There's no longer a `mono_glue=yes/no` SCons options. We no longer
need to build with `mono_glue=no`, generate the glue and then build
again with `mono_glue=yes`. We build only once and generate the glue
(which is in C# now).
However, SCons no longer builds the C# projects for us. Instead one
must run `build_assemblies.py`, e.g.:
```sh
%godot_src_root%/modules/mono/build_scripts/build_assemblies.py \
--godot-output-dir=%godot_src_root%/bin \
--godot-target=release_debug`
```
We could turn this into a custom build target, but I don't know how
to do that with SCons (it's possible with Meson).
OTHER NOTES
Most of the moved code doesn't follow the C# naming convention and
still has the word Mono in the names despite no longer dealing with
Mono's embedding APIs. This is just temporary while transitioning,
to make it easier to understand what was moved where.
2021-05-03 21:21:06 +08:00
|
|
|
};
|
2022-08-05 09:32:59 +08:00
|
|
|
|
|
|
|
const void **godotsharp::get_runtime_interop_funcs(int32_t &r_size) {
|
|
|
|
r_size = sizeof(unmanaged_callbacks);
|
|
|
|
return unmanaged_callbacks;
|
|
|
|
}
|