netcdf-c/ncgen/genc.c

1282 lines
35 KiB
C
Raw Normal View History

2010-06-03 21:24:43 +08:00
/*********************************************************************
2018-12-07 06:40:43 +08:00
* Copyright 2018, UCAR/Unidata
2010-06-03 21:24:43 +08:00
* See netcdf/COPYRIGHT file for copying and redistribution conditions.
* $Header: /upc/share/CVS/netcdf-3/ncgen/genc.c,v 1.6 2010/05/17 23:26:44 dmh Exp $
*********************************************************************/
#include "includes.h"
#include <ctype.h> /* for isprint() */
#ifdef ENABLE_C
#undef TRACE
/* Forward */
static const char* groupncid(Symbol*);
static const char* typencid(Symbol*);
static const char* varncid(Symbol*);
static const char* dimncid(Symbol*);
#ifdef USE_NETCDF4
static void definectype(Symbol*);
static void genc_deftype(Symbol*);
static void genc_definespecialattributes(Symbol* vsym);
static void genc_defineglobalspecials(void);
2010-06-03 21:24:43 +08:00
#endif
static void genc_defineattr(Symbol* asym);
static void genc_definevardata(Symbol*);
static void genc_write(Generator*,Symbol* sym, Bytebuffer* code,
int rank, size_t* start, size_t* count);
static void genc_writevar(Generator*,Symbol*,Bytebuffer*,int,size_t*,size_t*);
static void genc_writeattr(Generator*,Symbol*,Bytebuffer*,int,size_t*,size_t*);
2010-06-03 21:24:43 +08:00
/*
* Generate C code for creating netCDF from in-memory structure.
*/
void
Fix more memory leaks in netcdf-c library This is a follow up to PR https://github.com/Unidata/netcdf-c/pull/1173 Sorry that it is so big, but leak suppression can be complex. This PR fixes all remaining memory leaks -- as determined by -fsanitize=address, and with the exceptions noted below. Unfortunately. there remains a significant leak that I cannot solve. It involves vlens, and it is unclear if the leak is occurring in the netcdf-c library or the HDF5 library. I have added a check_PROGRAM to the ncdump directory to show the problem. The program is called tst_vlen_demo.c To exercise it, build the netcdf library with -fsanitize=address enabled. Then go into ncdump and do a "make clean check". This should build tst_vlen_demo without actually executing it. Then do the command "./tst_vlen_demo" to see the output of the memory checker. Note the the lost malloc is deep in the HDF5 library (in H5Tvlen.c). I am temporarily working around this error in the following way. 1. I modified several test scripts to not execute known vlen tests that fail as described above. 2. Added an environment variable called NC_VLEN_NOTEST. If set, then those specific tests are suppressed. This should mean that the --disable-utilities option to ./configure should not need to be set to get a memory leak clean build. This should allow for detection of any new leaks. Note: I used an environment variable rather than a ./configure option to control the vlen tests. This is because it is temporary (I hope) and because it is a bit tricky for shell scripts to access ./configure options. Finally, as before, this only been tested with netcdf-4 and hdf5 support.
2018-11-16 01:00:38 +08:00
genc_netcdf(void)
2010-06-03 21:24:43 +08:00
{
int idim, ivar, iatt, maxdims;
int ndims, nvars, natts, ngatts;
2010-06-03 21:24:43 +08:00
char* cmode_string;
Fix more memory leaks in netcdf-c library This is a follow up to PR https://github.com/Unidata/netcdf-c/pull/1173 Sorry that it is so big, but leak suppression can be complex. This PR fixes all remaining memory leaks -- as determined by -fsanitize=address, and with the exceptions noted below. Unfortunately. there remains a significant leak that I cannot solve. It involves vlens, and it is unclear if the leak is occurring in the netcdf-c library or the HDF5 library. I have added a check_PROGRAM to the ncdump directory to show the problem. The program is called tst_vlen_demo.c To exercise it, build the netcdf library with -fsanitize=address enabled. Then go into ncdump and do a "make clean check". This should build tst_vlen_demo without actually executing it. Then do the command "./tst_vlen_demo" to see the output of the memory checker. Note the the lost malloc is deep in the HDF5 library (in H5Tvlen.c). I am temporarily working around this error in the following way. 1. I modified several test scripts to not execute known vlen tests that fail as described above. 2. Added an environment variable called NC_VLEN_NOTEST. If set, then those specific tests are suppressed. This should mean that the --disable-utilities option to ./configure should not need to be set to get a memory leak clean build. This should allow for detection of any new leaks. Note: I used an environment variable rather than a ./configure option to control the vlen tests. This is because it is temporary (I hope) and because it is a bit tricky for shell scripts to access ./configure options. Finally, as before, this only been tested with netcdf-4 and hdf5 support.
2018-11-16 01:00:38 +08:00
const char *filename = rootgroup->file.filename;
2010-06-03 21:24:43 +08:00
#ifdef USE_NETCDF4
int igrp,ityp, ngrps, ntyps;
2010-06-03 21:24:43 +08:00
#endif
ndims = listlength(dimdefs);
nvars = listlength(vardefs);
natts = listlength(attdefs);
ngatts = listlength(gattdefs);
#ifdef USE_NETCDF4
2010-06-03 21:24:43 +08:00
ngrps = listlength(grpdefs);
ntyps = listlength(typdefs);
#endif /*USE_NETCDF4*/
2010-06-03 21:24:43 +08:00
/* wrap in main program */
codeline("#include <stdio.h>");
codeline("#include <stdlib.h>");
codeline("#include <netcdf.h>");
codeline("");
codeflush();
if(specialconstants) {
/* If the input referenced e.g. nan, inf, etc;
then provide definitions for them */
codeline("");
codeline("#define nanf (0.0f/0.0f)");
codeline("#define nan (0.0/0.0)");
codeline("#define inff (1.0f/0.0f)");
codeline("#define inf (1.0/0.0)");
codeline("#define infinityf inff");
codeline("#define infinity inf");
codeline("");
codeflush();
}
codeline("");
codeflush();
2010-06-03 21:24:43 +08:00
#ifdef USE_NETCDF4
/* Construct C type definitions*/
if (ntyps > 0) {
for(ityp = 0; ityp < ntyps; ityp++) {
Symbol* tsym = (Symbol*)listget(typdefs,ityp);
definectype(tsym);
}
codeline("");
}
codeflush();
/* Construct the chunking constants*/
if(!usingclassic) {
for(ivar=0;ivar<nvars;ivar++) {
Bytebuffer* tmp = bbNew();
Symbol* var = (Symbol*)listget(vardefs,ivar);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
Specialdata* special = &var->var.special;
2010-06-03 21:24:43 +08:00
if(special->flags & _CHUNKSIZES_FLAG) {
int i;
size_t* chunks = special->_ChunkSizes;
if(special->nchunks == 0 || chunks == NULL) continue;
bbClear(tmp);
for(i=0;i<special->nchunks;i++) {
bbprintf(tmp,"%s%ld",
(i == 0?"":", "),special->_ChunkSizes[i]);
}
bbprintf0(stmt,"static size_t %s_chunksizes[%d] = {",
cname(var),special->nchunks);
codedump(stmt);
codedump(tmp);
codeline("} ;");
}
if(special->flags & _FILTER_FLAG) {
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
int k;
for(k=0;k<special->nfilters;k++) {
int i;
size_t nparams;
Mostly revert the filter code to reduce its complexity of use. re: https://github.com/Unidata/netcdf-c/issues/1836 Revert the internal filter code to simplify it. From the user's point of view, the only visible changes should be: 1. The functions that convert text to filter specs have had their signature reverted and have been moved to netcdf_aux.h 2. Some filter API functions now return NC_ENOFILTER when inquiry is made about some filter. Internally,the dispatch table has been modified to get rid of the filter_actions entry and associated complex structures. It has been replaced with inq_var_filter_ids and inq_var_filter_info entries and the dispatch table version has been bumped to 3. Corresponding NOOP and NOTNC4 functions were added to libdispatch/dnotnc4.c. Also, the filter_action entries in dispatch tables were replaced for all dispatch code bases (HDF5, DAP2, etc). This should only impact UDF users. In the process, it became clear that the form of the filters field in NC_VAR_INFO_T was format dependent, so I converted it to be of type void* and pushed its management into the various dispatch code bases. Specifically libhdf5 and libnczarr now manage the filters field in their own way. The auxilliary functions for parsing textual filter specifications were moved to netcdf_aux.h and were renamed to the following: * ncaux_h5filterspec_parse * ncaux_h5filterspec_parselist * ncaux_h5filterspec_free * ncaux_h5filter_fix8 Misc. Other Changes: 1. Document NUG/filters.md updated to reflect the changes above. 2. All the old data types (structs and enums) used by filter_actions actions were deleted. The exception is the NC_H5_Filterspec because it is needed by ncaux_h5filterspec_parselist. 3. Clientside filters were removed -- another enhancement for which no-one ever asked. 4. The ability to remove filters was itself removed. 5. Some functionality needed by nczarr was moved from libhdf5 to libsrc4 e.g. nc4_find_default_chunksizes 6. All the filterx code was removed 7. ncfilter.h and nc4filter.c no longer used Misc. Unrelated Changes: 1. The nczarr_test makefile clean was leaving some directories; so add clean-local to take care of them.
2020-09-28 02:43:46 +08:00
unsigned int* params = NULL;
NC_H5_Filterspec* nfs = special->_Filters[k];
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
if(nfs->nparams == 0 || nfs->params == NULL) continue;
bbClear(tmp);
nparams = nfs->nparams;
params = nfs->params;
for(i=0;i<nparams;i++) {
Mostly revert the filter code to reduce its complexity of use. re: https://github.com/Unidata/netcdf-c/issues/1836 Revert the internal filter code to simplify it. From the user's point of view, the only visible changes should be: 1. The functions that convert text to filter specs have had their signature reverted and have been moved to netcdf_aux.h 2. Some filter API functions now return NC_ENOFILTER when inquiry is made about some filter. Internally,the dispatch table has been modified to get rid of the filter_actions entry and associated complex structures. It has been replaced with inq_var_filter_ids and inq_var_filter_info entries and the dispatch table version has been bumped to 3. Corresponding NOOP and NOTNC4 functions were added to libdispatch/dnotnc4.c. Also, the filter_action entries in dispatch tables were replaced for all dispatch code bases (HDF5, DAP2, etc). This should only impact UDF users. In the process, it became clear that the form of the filters field in NC_VAR_INFO_T was format dependent, so I converted it to be of type void* and pushed its management into the various dispatch code bases. Specifically libhdf5 and libnczarr now manage the filters field in their own way. The auxilliary functions for parsing textual filter specifications were moved to netcdf_aux.h and were renamed to the following: * ncaux_h5filterspec_parse * ncaux_h5filterspec_parselist * ncaux_h5filterspec_free * ncaux_h5filter_fix8 Misc. Other Changes: 1. Document NUG/filters.md updated to reflect the changes above. 2. All the old data types (structs and enums) used by filter_actions actions were deleted. The exception is the NC_H5_Filterspec because it is needed by ncaux_h5filterspec_parselist. 3. Clientside filters were removed -- another enhancement for which no-one ever asked. 4. The ability to remove filters was itself removed. 5. Some functionality needed by nczarr was moved from libhdf5 to libsrc4 e.g. nc4_find_default_chunksizes 6. All the filterx code was removed 7. ncfilter.h and nc4filter.c no longer used Misc. Unrelated Changes: 1. The nczarr_test makefile clean was leaving some directories; so add clean-local to take care of them.
2020-09-28 02:43:46 +08:00
bbprintf(tmp,"%s%uU",
(i == 0?"":", "),params[i]);
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
}
bbprintf0(stmt,"static unsigned int %s_%d_filterparams[%d] = {",
cname(var),k,nparams);
codedump(stmt);
codedump(tmp);
codeline("} ;");
}
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
}
bbFree(tmp);
2010-06-03 21:24:43 +08:00
}
codeline("");
}
#endif /*USE_NETCDF4*/
/* Now construct the main procedures*/
codeline("void");
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codeline("check_err(const int stat, int line, const char* file, const char* func) {");
2010-06-03 21:24:43 +08:00
codelined(1,"if (stat != NC_NOERR) {");
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(2,"(void)fprintf(stderr,\"line %d of %s.%s: %s\\n\", line, file, func, nc_strerror(stat));");
2010-06-03 21:24:43 +08:00
codelined(2,"fflush(stderr);");
codelined(2,"exit(1);");
codelined(1,"}");
codeline("}");
codeline("");
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codeline("#define CHECK_ERR(err) check_err(err, __LINE__, __FILE__, __func__)");
codeline("");
2010-06-03 21:24:43 +08:00
codeline("int");
bbprintf0(stmt,"%s() {/* create %s */\n", mainname, filename);
codedump(stmt);
/* create necessary declarations */
codeline("");
2010-06-03 21:24:43 +08:00
codelined(1,"int stat; /* return status */");
codelined(1,"int ncid; /* netCDF id */");
codeflush();
#ifdef USE_NETCDF4
/* Define variables to hold group ids*/
if(!usingclassic) {
codeline("");
codelined(1,"/* group ids */");
}
if(!usingclassic && ngrps > 0) {
2010-06-03 21:24:43 +08:00
for(igrp = 0; igrp < ngrps; igrp++) {
Symbol* gsym = (Symbol*)listget(grpdefs,igrp);
bbprintf0(stmt,"%sint %s;\n",indented(1),groupncid(gsym));
codedump(stmt);
}
}
/* define variables to hold type ids*/
if(!usingclassic && ntyps > 0) {
codeline("");
codelined(1,"/* type ids */");
for(ityp = 0; ityp < ntyps; ityp++) {
Symbol* tsym = (Symbol*)listget(typdefs,ityp);
bbprintf0(stmt,"%sint %s;\n",indented(1), typencid(tsym));
codedump(stmt);
}
}
codeflush();
#endif
if (ndims > 0) {
codeline("");
codelined(1,"/* dimension ids */");
for(idim = 0; idim < ndims; idim++) {
Symbol* dsym = (Symbol*)listget(dimdefs,idim);
bbprintf0(stmt,"%sint %s;\n", indented(1), dimncid(dsym));
codedump(stmt);
}
codeline("");
codelined(1,"/* dimension lengths */");
for(idim = 0; idim < ndims; idim++) {
Symbol* dsym = (Symbol*)listget(dimdefs,idim);
if(dsym->dim.isunlimited) {
2010-06-03 21:24:43 +08:00
bbprintf0(stmt,"%ssize_t %s_len = NC_UNLIMITED;\n",
indented(1),cname(dsym));
} else {
bbprintf0(stmt,"%ssize_t %s_len = %lu;\n",
indented(1),
cname(dsym),
(unsigned long) dsym->dim.declsize);
}
codedump(stmt);
}
}
codeflush();
maxdims = 0; /* most dimensions of any variable */
for(ivar = 0; ivar < nvars; ivar++) {
Symbol* vsym = (Symbol*)listget(vardefs,ivar);
if(vsym->typ.dimset.ndims > maxdims)
maxdims = vsym->typ.dimset.ndims;
}
if (nvars > 0) {
codeline("");
codelined(1,"/* variable ids */");
for(ivar = 0; ivar < nvars; ivar++) {
Symbol* vsym = (Symbol*)listget(vardefs,ivar);
bbprintf0(stmt," int %s;\n", varncid(vsym));
codedump(stmt);
}
codeline("");
codelined(1,"/* rank (number of dimensions) for each variable */");
for(ivar = 0; ivar < nvars; ivar++) {
Symbol* vsym = (Symbol*)listget(vardefs,ivar);
bbprintf0(stmt,"# define RANK_%s %d\n", cname(vsym),
vsym->typ.dimset.ndims);
codedump(stmt);
}
if (maxdims > 0) { /* we have dimensioned variables */
codeline("");
codelined(1,"/* variable shapes */");
for(ivar = 0; ivar < nvars; ivar++) {
Symbol* vsym = (Symbol*)listget(vardefs,ivar);
if (vsym->typ.dimset.ndims > 0) {
bbprintf0(stmt," int %s_dims[RANK_%s];\n",
cname(vsym), cname(vsym));
codedump(stmt);
}
}
}
}
codeflush();
/* Set log level */
if(ncloglevel >= 0) {
codeline("");
bbprintf0(stmt," nc_set_log_level(%d); /* set log level */",ncloglevel);
codedump(stmt);
codeline("");
}
2010-06-03 21:24:43 +08:00
/* create netCDF file, uses NC_CLOBBER mode */
codeline("");
codelined(1,"/* enter define mode */");
if (!cmode_modifier) {
cmode_string = "NC_CLOBBER";
} else if (cmode_modifier & NC_64BIT_OFFSET) {
cmode_string = "NC_CLOBBER|NC_64BIT_OFFSET";
#ifdef USE_NETCDF4
} else if (cmode_modifier & NC_CLASSIC_MODEL) {
cmode_string = "NC_CLOBBER|NC_NETCDF4|NC_CLASSIC_MODEL";
} else if (cmode_modifier & NC_NETCDF4) {
cmode_string = "NC_CLOBBER|NC_NETCDF4";
2010-06-03 21:24:43 +08:00
#endif
} else {
derror("unknown cmode modifier");
cmode_string = "NC_CLOBBER";
}
bbprintf0(stmt," stat = nc_create(\"%s\", %s, &ncid);\n",
filename,cmode_string);
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
codeflush();
#ifdef USE_NETCDF4
genc_defineglobalspecials();
#endif /*USE_NETCDF4*/
2010-06-03 21:24:43 +08:00
#ifdef USE_NETCDF4
/* Define the group structure */
/* ncid created above is also root group*/
if(!usingclassic) {
bbprintf0(stmt," %s = ncid;\n",groupncid(rootgroup));
codedump(stmt);
/* walking grpdefs list will do a preorder walk of all defined groups*/
for(igrp=0;igrp<listlength(grpdefs);igrp++) {
Symbol* gsym = (Symbol*)listget(grpdefs,igrp);
if(gsym == rootgroup) continue; /* ignore root*/
if(gsym->container == NULL)
PANIC("null container");
bbprintf0(stmt,
" stat = nc_def_grp(%s, \"%s\", &%s);\n",
groupncid(gsym->container),
gsym->name, groupncid(gsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
}
codeflush();
}
#endif
#ifdef USE_NETCDF4
/* Construct code to define types*/
if(ntyps > 0) {
codeline("");
for(ityp = 0; ityp < ntyps; ityp++) {
Symbol* tsym = (Symbol*)listget(typdefs,ityp);
if(tsym->subclass == NC_PRIM
|| tsym->subclass == NC_ARRAY) continue; /* no need to do these*/
genc_deftype(tsym);
codeline("");
}
}
codeflush();
#endif
/* define dimensions from info in dims array */
if (ndims > 0) {
codeline("");
codelined(1,"/* define dimensions */");
for(idim = 0; idim < ndims; idim++) {
Symbol* dsym = (Symbol*)listget(dimdefs,idim);
{
bbprintf0(stmt,
" stat = nc_def_dim(%s, \"%s\", %s_len, &%s);\n",
groupncid(dsym->container),
escapifyname(dsym->name),
cname(dsym),
dimncid(dsym));
}
2010-06-03 21:24:43 +08:00
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
}
}
codeflush();
/* define variables from info in vars array */
if (nvars > 0) {
codeline("");
codelined(1,"/* define variables */");
for(ivar = 0; ivar < nvars; ivar++) {
Symbol* vsym = (Symbol*)listget(vardefs,ivar);
Symbol* basetype = vsym->typ.basetype;
Dimset* dimset = &vsym->typ.dimset;
codeline("");
if(dimset->ndims > 0) {
for(idim = 0; idim < dimset->ndims; idim++) {
Symbol* dsym = dimset->dimsyms[idim];
bbprintf0(stmt,
" %s_dims[%d] = %s;\n",
cname(vsym),
idim,
dimncid(dsym));
codedump(stmt);
}
}
bbprintf0(stmt,
" stat = nc_def_var(%s, \"%s\", %s, RANK_%s, %s, &%s);\n",
groupncid(vsym->container),
escapifyname(vsym->name),
typencid(basetype),
cname(vsym),
(dimset->ndims == 0?"0":poolcat(cname(vsym),"_dims")),
varncid(vsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
#ifdef USE_NETCDF4
genc_definespecialattributes(vsym);
#endif /*USE_NETCDF4*/
}
}
codeflush();
2010-06-03 21:24:43 +08:00
/* Define the global attributes*/
if(ngatts > 0) {
codeline("");
codelined(1,"/* assign global attributes */");
for(iatt = 0; iatt < ngatts; iatt++) {
Symbol* gasym = (Symbol*)listget(gattdefs,iatt);
genc_defineattr(gasym);
2010-06-03 21:24:43 +08:00
}
codeline("");
}
codeflush();
2010-06-03 21:24:43 +08:00
/* Define the variable specific attributes*/
if(natts > 0) {
codeline("");
codelined(1,"/* assign per-variable attributes */");
for(iatt = 0; iatt < natts; iatt++) {
Symbol* asym = (Symbol*)listget(attdefs,iatt);
genc_defineattr(asym);
}
codeline("");
}
codeflush();
if (nofill_flag) {
codelined(1,"/* don't initialize variables with fill values */");
bbindent(stmt,1);
bbprintf0(stmt,"stat = nc_set_fill(%s, NC_NOFILL, 0);",groupncid(rootgroup));
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
}
codeline("");
codelined(1,"/* leave define mode */");
bbprintf0(stmt," stat = nc_enddef (%s);\n",groupncid(rootgroup));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
codeflush();
2012-03-08 07:38:51 +08:00
if(!header_only) {
/* Load values into those variables with defined data */
if(nvars > 0) {
codeline("");
codelined(1,"/* assign variable data */");
for(ivar = 0; ivar < nvars; ivar++) {
Symbol* vsym = (Symbol*)listget(vardefs,ivar);
if(vsym->data != NULL) genc_definevardata(vsym);
}
codeline("");
}
codeflush();
2010-06-03 21:24:43 +08:00
}
}
#ifdef USE_NETCDF4
static void
genc_defineglobalspecials(void)
{
if(usingclassic) return;
if(!/*Main.*/format_attribute) return;
/* There are currently no global specials that
can be defined using nc_put_att.*/
}
2010-06-03 21:24:43 +08:00
static void
genc_definespecialattributes(Symbol* vsym)
{
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
Specialdata* special = &vsym->var.special;
2010-06-03 21:24:43 +08:00
if(usingclassic) return;
if(special->flags & _STORAGE_FLAG) {
const char* storage = NULL;
2010-06-03 21:24:43 +08:00
size_t* chunks = special->_ChunkSizes;
switch (special->_Storage) {
case NC_CONTIGUOUS: storage = "NC_CONTIGUOUS"; break;
case NC_COMPACT: storage = "NC_COMPACT"; break;
case NC_CHUNKED: storage = "NC_CHUNKED"; break;
}
2010-06-03 21:24:43 +08:00
bbprintf0(stmt,
" stat = nc_def_var_chunking(%s, %s, %s, ",
groupncid(vsym->container),
varncid(vsym),
storage);
2010-06-03 21:24:43 +08:00
codedump(stmt);
if(special->nchunks == 0 || chunks == NULL)
codepartial("NULL");
else {
bbprintf0(stmt,"%s_chunksizes",cname(vsym));
codedump(stmt);
2010-06-03 21:24:43 +08:00
}
codeline(");");
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
}
2010-06-03 21:24:43 +08:00
if(special->flags & _FLETCHER32_FLAG) {
bbprintf0(stmt,
" stat = nc_def_var_fletcher32(%s, %s, %d);\n",
groupncid(vsym->container),
varncid(vsym),
special->_Fletcher32);
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
}
if(special->flags & (_DEFLATE_FLAG | _SHUFFLE_FLAG)) {
bbprintf0(stmt,
" stat = nc_def_var_deflate(%s, %s, %s, %d, %d);\n",
groupncid(vsym->container),
varncid(vsym),
(special->_Shuffle == 1?"NC_SHUFFLE":"NC_NOSHUFFLE"),
(special->_DeflateLevel >= 0?1:0),
(special->_DeflateLevel >= 0?special->_DeflateLevel:0));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
}
2010-06-03 21:24:43 +08:00
if(special->flags & _ENDIAN_FLAG) {
bbprintf0(stmt,
" stat = nc_def_var_endian(%s, %s, %s);\n",
groupncid(vsym->container),
varncid(vsym),
(special->_Endianness == NC_ENDIAN_LITTLE?"NC_ENDIAN_LITTLE"
:"NC_ENDIAN_BIG")
);
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
}
2010-06-03 21:24:43 +08:00
if(special->flags & _NOFILL_FLAG) {
bbprintf0(stmt,
" stat = nc_def_var_fill(%s, %s, %s, NULL);\n",
groupncid(vsym->container),
varncid(vsym),
(special->_Fill?"NC_FILL":"NC_NOFILL")
);
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
}
if(special->flags & _FILTER_FLAG) {
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
int k;
for(k=0;k<special->nfilters;k++) {
Mostly revert the filter code to reduce its complexity of use. re: https://github.com/Unidata/netcdf-c/issues/1836 Revert the internal filter code to simplify it. From the user's point of view, the only visible changes should be: 1. The functions that convert text to filter specs have had their signature reverted and have been moved to netcdf_aux.h 2. Some filter API functions now return NC_ENOFILTER when inquiry is made about some filter. Internally,the dispatch table has been modified to get rid of the filter_actions entry and associated complex structures. It has been replaced with inq_var_filter_ids and inq_var_filter_info entries and the dispatch table version has been bumped to 3. Corresponding NOOP and NOTNC4 functions were added to libdispatch/dnotnc4.c. Also, the filter_action entries in dispatch tables were replaced for all dispatch code bases (HDF5, DAP2, etc). This should only impact UDF users. In the process, it became clear that the form of the filters field in NC_VAR_INFO_T was format dependent, so I converted it to be of type void* and pushed its management into the various dispatch code bases. Specifically libhdf5 and libnczarr now manage the filters field in their own way. The auxilliary functions for parsing textual filter specifications were moved to netcdf_aux.h and were renamed to the following: * ncaux_h5filterspec_parse * ncaux_h5filterspec_parselist * ncaux_h5filterspec_free * ncaux_h5filter_fix8 Misc. Other Changes: 1. Document NUG/filters.md updated to reflect the changes above. 2. All the old data types (structs and enums) used by filter_actions actions were deleted. The exception is the NC_H5_Filterspec because it is needed by ncaux_h5filterspec_parselist. 3. Clientside filters were removed -- another enhancement for which no-one ever asked. 4. The ability to remove filters was itself removed. 5. Some functionality needed by nczarr was moved from libhdf5 to libsrc4 e.g. nc4_find_default_chunksizes 6. All the filterx code was removed 7. ncfilter.h and nc4filter.c no longer used Misc. Unrelated Changes: 1. The nczarr_test makefile clean was leaving some directories; so add clean-local to take care of them.
2020-09-28 02:43:46 +08:00
NC_H5_Filterspec* nfs = special->_Filters[k];
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
bbprintf0(stmt,
Mostly revert the filter code to reduce its complexity of use. re: https://github.com/Unidata/netcdf-c/issues/1836 Revert the internal filter code to simplify it. From the user's point of view, the only visible changes should be: 1. The functions that convert text to filter specs have had their signature reverted and have been moved to netcdf_aux.h 2. Some filter API functions now return NC_ENOFILTER when inquiry is made about some filter. Internally,the dispatch table has been modified to get rid of the filter_actions entry and associated complex structures. It has been replaced with inq_var_filter_ids and inq_var_filter_info entries and the dispatch table version has been bumped to 3. Corresponding NOOP and NOTNC4 functions were added to libdispatch/dnotnc4.c. Also, the filter_action entries in dispatch tables were replaced for all dispatch code bases (HDF5, DAP2, etc). This should only impact UDF users. In the process, it became clear that the form of the filters field in NC_VAR_INFO_T was format dependent, so I converted it to be of type void* and pushed its management into the various dispatch code bases. Specifically libhdf5 and libnczarr now manage the filters field in their own way. The auxilliary functions for parsing textual filter specifications were moved to netcdf_aux.h and were renamed to the following: * ncaux_h5filterspec_parse * ncaux_h5filterspec_parselist * ncaux_h5filterspec_free * ncaux_h5filter_fix8 Misc. Other Changes: 1. Document NUG/filters.md updated to reflect the changes above. 2. All the old data types (structs and enums) used by filter_actions actions were deleted. The exception is the NC_H5_Filterspec because it is needed by ncaux_h5filterspec_parselist. 3. Clientside filters were removed -- another enhancement for which no-one ever asked. 4. The ability to remove filters was itself removed. 5. Some functionality needed by nczarr was moved from libhdf5 to libsrc4 e.g. nc4_find_default_chunksizes 6. All the filterx code was removed 7. ncfilter.h and nc4filter.c no longer used Misc. Unrelated Changes: 1. The nczarr_test makefile clean was leaving some directories; so add clean-local to take care of them.
2020-09-28 02:43:46 +08:00
" stat = nc_def_var_filter(%s, %s, %u, %lu, ",
groupncid(vsym->container),
varncid(vsym),
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
nfs->filterid,
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
(unsigned long)nfs->nparams
);
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
codedump(stmt);
if(nfs->nparams == 0 || nfs->params == NULL)
codepartial("NULL");
else {
bbprintf0(stmt,"%s_%d_filterparams",cname(vsym),k);
codedump(stmt);
Add support for multiple filters per variable. re: https://github.com/Unidata/netcdf-c/issues/1584 Support has been added for multiple filters per variable. This affects a number of components in netcdf. The new APIs are documented in NUG/filters.md. The primary changes are: * A set of new functions are provided (see __include/netcdf_filter.h__). - Obtain a list of the filters associated with a variable - Obtain the parameters for a specific filter. * The existing __nc_inq_var_filter__ function now returns info about the first defined filter. * The utilities (ncgen, ncdump, and nccopy) now support an extended format for specifying a sequence of filters. The general form is __<filter>|<filter>..._. * The ncdump **_Filter** attribute now dumps a list of all the filters associated with a variable using the above new format. * Filter specifications can now use a filter name instead of number for filters known to the netcdf library, which in turn is taken from the HDF5 filter registration page. * New errors are defined: NC_EFILTER and NC_ENOFILTER. The latter is returned if an attempt is made to access an unknown filter. * Internally, the dispatch table has been extended to add a function to handle all of the filter functions. * New, filter-related, tests were added to nc_test4. * A new plugin was added to the plugins directory to help with testing. Notes: 1. The shuffle and fletcher32 filters are not part of the multifilter system. Misc. changes: 1. A debug module was added to libhdf5 to help catch error locations.
2020-02-17 03:59:33 +08:00
}
codeline(");");
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
}
}
2010-06-03 21:24:43 +08:00
}
#endif /*USE_NETCDF4*/
void
Fix more memory leaks in netcdf-c library This is a follow up to PR https://github.com/Unidata/netcdf-c/pull/1173 Sorry that it is so big, but leak suppression can be complex. This PR fixes all remaining memory leaks -- as determined by -fsanitize=address, and with the exceptions noted below. Unfortunately. there remains a significant leak that I cannot solve. It involves vlens, and it is unclear if the leak is occurring in the netcdf-c library or the HDF5 library. I have added a check_PROGRAM to the ncdump directory to show the problem. The program is called tst_vlen_demo.c To exercise it, build the netcdf library with -fsanitize=address enabled. Then go into ncdump and do a "make clean check". This should build tst_vlen_demo without actually executing it. Then do the command "./tst_vlen_demo" to see the output of the memory checker. Note the the lost malloc is deep in the HDF5 library (in H5Tvlen.c). I am temporarily working around this error in the following way. 1. I modified several test scripts to not execute known vlen tests that fail as described above. 2. Added an environment variable called NC_VLEN_NOTEST. If set, then those specific tests are suppressed. This should mean that the --disable-utilities option to ./configure should not need to be set to get a memory leak clean build. This should allow for detection of any new leaks. Note: I used an environment variable rather than a ./configure option to control the vlen tests. This is because it is temporary (I hope) and because it is a bit tricky for shell scripts to access ./configure options. Finally, as before, this only been tested with netcdf-4 and hdf5 support.
2018-11-16 01:00:38 +08:00
genc_close(void)
2010-06-03 21:24:43 +08:00
{
bbprintf0(stmt,"%sstat = nc_close(%s);\n",indented(1),groupncid(rootgroup));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
#ifndef vms
codelined(1,"return 0;");
#else
codelined(1,"return 1;");
#endif
codeline("}");
codeflush();
}
/*
* Output a C statement
*/
/* return C name for netCDF type, given type code */
const char *
nctype(nc_type type)
{
switch (type) {
case NC_CHAR: return "NC_CHAR";
case NC_BYTE: return "NC_BYTE";
case NC_SHORT: return "NC_SHORT";
case NC_INT: return "NC_INT";
case NC_FLOAT: return "NC_FLOAT";
case NC_DOUBLE: return "NC_DOUBLE";
case NC_UBYTE: return "NC_UBYTE";
case NC_USHORT: return "NC_USHORT";
case NC_UINT: return "NC_UINT";
case NC_INT64: return "NC_INT64";
case NC_UINT64: return "NC_UINT64";
case NC_STRING: return "NC_STRING";
default: PANIC("nctype: bad type code");
}
return NULL;
}
/*
* Return C type name for netCDF type, given type code.
*/
const char*
2010-06-03 21:24:43 +08:00
ncctype(nc_type type)
{
switch (type) {
case NC_CHAR:
return "char";
case NC_BYTE:
return "signed char";
case NC_SHORT:
return "short";
case NC_INT:
return "int";
case NC_FLOAT:
return "float";
case NC_DOUBLE:
return "double";
case NC_UBYTE:
return "unsigned char";
case NC_USHORT:
return "unsigned short";
case NC_UINT:
return "unsigned int";
case NC_INT64:
return "signed long long";
case NC_UINT64:
return "unsigned long long";
case NC_STRING:
return "char*";
default:
PANIC1("ncctype: bad type code:%d",type);
}
return 0;
}
/*
* Return C type name for netCDF type suffix, given type code.
*/
const char*
2010-06-03 21:24:43 +08:00
ncstype(nc_type nctype)
{
switch (nctype) {
case NC_CHAR:
return "text";
case NC_BYTE:
return "schar";
case NC_SHORT:
return "short";
case NC_INT:
return "int";
case NC_FLOAT:
return "float";
case NC_DOUBLE:
return "double";
case NC_UBYTE:
return "ubyte";
case NC_USHORT:
return "ushort";
case NC_UINT:
return "uint";
case NC_INT64:
return "longlong";
case NC_UINT64:
return "ulonglong";
case NC_STRING:
return "string";
default:
derror("ncstype: bad type code: %d",nctype);
return 0;
}
}
/* Return the group name for the specified group*/
static const char*
groupncid(Symbol* sym)
{
#ifdef USE_NETCDF4
if(usingclassic) {
return "ncid";
} else {
char* grptmp;
const char* tmp1;
if(sym == NULL) return groupncid(rootgroup);
ASSERT(sym->objectclass == NC_GRP);
tmp1 = cname(sym);
grptmp = poolalloc(strlen(tmp1)+strlen("_grp")+1);
strcpy(grptmp,tmp1);
strcat(grptmp,"_grp");
return grptmp;
}
#else
return "ncid";
#endif
}
/* Compute the C name for a given type's id*/
/* Watch out: the result is a static*/
static const char*
typencid(Symbol* tsym)
{
char* typtmp;
const char* tmp1;
if(tsym->subclass == NC_PRIM)
return nctype(tsym->typ.typecode);
tmp1 = ctypename(tsym);
typtmp = poolalloc(strlen(tmp1)+strlen("_typ")+1);
strcpy(typtmp,tmp1);
strcat(typtmp,"_typ");
return typtmp;
}
/* Compute the C name for a given var's id*/
/* Watch out: the result is a static*/
static const char*
varncid(Symbol* vsym)
{
const char* tmp1;
char* vartmp;
tmp1 = cname(vsym);
vartmp = poolalloc(strlen(tmp1)+strlen("_id")+1);
strcpy(vartmp,tmp1);
strcat(vartmp,"_id");
return vartmp;
}
/* Compute the C name for a given dim's id*/
/* Watch out: the result is pooled*/
2010-06-03 21:24:43 +08:00
static const char*
dimncid(Symbol* dsym)
{
const char* tmp1;
char* dimtmp;
tmp1 = cname(dsym);
dimtmp = poolalloc(strlen(tmp1)+strlen("_dim")+1);
2010-06-03 21:24:43 +08:00
strcpy(dimtmp,tmp1);
strcat(dimtmp,"_dim");
2010-06-03 21:24:43 +08:00
return dimtmp;
}
/* Compute the C name for a given type*/
const char*
ctypename(Symbol* tsym)
{
const char* name;
ASSERT(tsym->objectclass == NC_TYPE);
if(tsym->subclass == NC_PRIM)
name = ncctype(tsym->typ.typecode);
else
name = cname(tsym);
return name;
}
#ifdef USE_NETCDF4
static void
definectype(Symbol* tsym)
{
int i,j;
ASSERT(tsym->objectclass == NC_TYPE);
switch (tsym->subclass) {
case NC_PRIM: break; /* these are already taken care of*/
case NC_OPAQUE:
bbprintf0(stmt,"typedef unsigned char %s[%lu];\n",
cname(tsym), tsym->typ.size);
codedump(stmt);
break;
case NC_ENUM:
for(i=0;i<listlength(tsym->subnodes);i++) {
Symbol* econst = (Symbol*)listget(tsym->subnodes,i);
Bytebuffer* econststring = bbNew();
2010-06-03 21:24:43 +08:00
ASSERT(econst->subclass == NC_ECONST);
Fix more memory leaks in netcdf-c library This is a follow up to PR https://github.com/Unidata/netcdf-c/pull/1173 Sorry that it is so big, but leak suppression can be complex. This PR fixes all remaining memory leaks -- as determined by -fsanitize=address, and with the exceptions noted below. Unfortunately. there remains a significant leak that I cannot solve. It involves vlens, and it is unclear if the leak is occurring in the netcdf-c library or the HDF5 library. I have added a check_PROGRAM to the ncdump directory to show the problem. The program is called tst_vlen_demo.c To exercise it, build the netcdf library with -fsanitize=address enabled. Then go into ncdump and do a "make clean check". This should build tst_vlen_demo without actually executing it. Then do the command "./tst_vlen_demo" to see the output of the memory checker. Note the the lost malloc is deep in the HDF5 library (in H5Tvlen.c). I am temporarily working around this error in the following way. 1. I modified several test scripts to not execute known vlen tests that fail as described above. 2. Added an environment variable called NC_VLEN_NOTEST. If set, then those specific tests are suppressed. This should mean that the --disable-utilities option to ./configure should not need to be set to get a memory leak clean build. This should allow for detection of any new leaks. Note: I used an environment variable rather than a ./configure option to control the vlen tests. This is because it is temporary (I hope) and because it is a bit tricky for shell scripts to access ./configure options. Finally, as before, this only been tested with netcdf-4 and hdf5 support.
2018-11-16 01:00:38 +08:00
c_generator->constant(c_generator,tsym,econst->typ.econst,econststring);
bbNull(econststring);
/* Enum constants must be converted to a fully qualified name */
2010-06-03 21:24:43 +08:00
bbprintf0(stmt,"#define %s ((%s)%s)\n",
cname(econst),
ctypename(econst->typ.basetype),
bbContents(econststring));
bbFree(econststring);
2010-06-03 21:24:43 +08:00
codedump(stmt);
}
bbprintf0(stmt,"typedef %s %s;\n",
ctypename(tsym->typ.basetype), cname(tsym));
codedump(stmt);
break;
case NC_VLEN:
bbprintf0(stmt,"typedef nc_vlen_t %s;\n",
ctypename(tsym));
codedump(stmt);
break;
case NC_COMPOUND:
bbprintf0(stmt,"typedef struct %s {\n",cname(tsym));
codedump(stmt);
for(i=0;i<listlength(tsym->subnodes);i++) {
Symbol* efield = (Symbol*)listget(tsym->subnodes,i);
ASSERT(efield->subclass == NC_FIELD);
bbprintf0(stmt,"%s%s %s",
indented(1),ctypename(efield->typ.basetype),cname(efield));
codedump(stmt);
/* compute any dimension specification*/
if(efield->typ.dimset.ndims > 0) {
Bytebuffer* dimbuf = bbNew();
for(j=0;j<efield->typ.dimset.ndims;j++) {
Symbol* dim;
char tmp[64];
bbCat(dimbuf,"[");
dim = efield->typ.dimset.dimsyms[j];
ASSERT(dim->dim.isconstant);
snprintf(tmp,sizeof(tmp),"%u",
(unsigned int)dim->dim.declsize);
bbCat(dimbuf,tmp);
bbCat(dimbuf,"]");
}
codedump(dimbuf);
bbFree(dimbuf);
}
codeline(";");
}
bbprintf0(stmt,"} %s;\n", ctypename(tsym));
codedump(stmt);
break;
case NC_ARRAY:
/* ignore: this will be handled by def_var*/
break;
default: panic("definectype: unexpected type subclass: %d",tsym->subclass);
}
}
/*
Generate the C code for defining a given type
*/
static void
genc_deftype(Symbol* tsym)
{
int i;
ASSERT(tsym->objectclass == NC_TYPE);
switch (tsym->subclass) {
case NC_PRIM: break; /* these are already taken care of*/
case NC_OPAQUE:
bbprintf0(stmt,"%sstat = nc_def_opaque(%s, %lu, \"%s\", &%s);\n",
indented(1),
groupncid(tsym->container),
tsym->typ.size,
tsym->name,
typencid(tsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
break;
case NC_ENUM:
codelined(1,"{");
bbprintf0(stmt,"%s%s econst;\n",
indented(1),
ncctype(tsym->typ.basetype->typ.typecode));
codedump(stmt);
bbprintf0(stmt,"%sstat = nc_def_enum(%s, %s, \"%s\", &%s);\n",
indented(1),
groupncid(tsym->container),
nctype(tsym->typ.basetype->typ.typecode),
tsym->name,
typencid(tsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
for(i=0;i<listlength(tsym->subnodes);i++) {
Symbol* econst = (Symbol*)listget(tsym->subnodes,i);
Bytebuffer* econststring = bbNew();
2010-06-03 21:24:43 +08:00
ASSERT(econst->subclass == NC_ECONST);
Fix more memory leaks in netcdf-c library This is a follow up to PR https://github.com/Unidata/netcdf-c/pull/1173 Sorry that it is so big, but leak suppression can be complex. This PR fixes all remaining memory leaks -- as determined by -fsanitize=address, and with the exceptions noted below. Unfortunately. there remains a significant leak that I cannot solve. It involves vlens, and it is unclear if the leak is occurring in the netcdf-c library or the HDF5 library. I have added a check_PROGRAM to the ncdump directory to show the problem. The program is called tst_vlen_demo.c To exercise it, build the netcdf library with -fsanitize=address enabled. Then go into ncdump and do a "make clean check". This should build tst_vlen_demo without actually executing it. Then do the command "./tst_vlen_demo" to see the output of the memory checker. Note the the lost malloc is deep in the HDF5 library (in H5Tvlen.c). I am temporarily working around this error in the following way. 1. I modified several test scripts to not execute known vlen tests that fail as described above. 2. Added an environment variable called NC_VLEN_NOTEST. If set, then those specific tests are suppressed. This should mean that the --disable-utilities option to ./configure should not need to be set to get a memory leak clean build. This should allow for detection of any new leaks. Note: I used an environment variable rather than a ./configure option to control the vlen tests. This is because it is temporary (I hope) and because it is a bit tricky for shell scripts to access ./configure options. Finally, as before, this only been tested with netcdf-4 and hdf5 support.
2018-11-16 01:00:38 +08:00
c_generator->constant(c_generator,tsym,econst->typ.econst,econststring);
bbNull(econststring);
2010-06-03 21:24:43 +08:00
bbprintf0(stmt,"%seconst = %s;\n",
indented(1),bbContents(econststring));
bbFree(econststring);
2010-06-03 21:24:43 +08:00
codedump(stmt);
bbprintf0(stmt,"%sstat = nc_insert_enum(%s, %s, \"%s\", &econst);\n",
indented(1),
groupncid(tsym->container),
typencid(tsym),
escapifyname(econst->name));
codedump(stmt);
}
codelined(1,"}");
break;
case NC_VLEN:
bbprintf0(stmt,"%sstat = nc_def_vlen(%s, \"%s\", %s, &%s);",
indented(1),
groupncid(tsym->container),
escapifyname(tsym->name),
typencid(tsym->typ.basetype),
typencid(tsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
break;
case NC_COMPOUND:
bbprintf0(stmt,"%sstat = nc_def_compound(%s, sizeof(%s), \"%s\", &%s);",
indented(1),
groupncid(tsym->container),
ctypename(tsym),
escapifyname(tsym->name),
typencid(tsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
/* Generate the field dimension constants*/
codelined(1,"{");
for(i=0;i<listlength(tsym->subnodes);i++) {
int j;
Symbol* efield = (Symbol*)listget(tsym->subnodes,i);
ASSERT(efield->subclass == NC_FIELD);
if(efield->typ.dimset.ndims == 0) continue;
2010-06-03 21:24:43 +08:00
bbprintf0(stmt,"%sstatic int %s_dims[%d] = {\n",
indented(1),
cname(efield),efield->typ.dimset.ndims);
for(j=0;j<efield->typ.dimset.ndims;j++) {
char tmp[256];
Symbol* e = efield->typ.dimset.dimsyms[j];
ASSERT(e->dim.isconstant);
snprintf(tmp,sizeof(tmp),"%u",(unsigned int)e->dim.declsize);
bbCat(stmt,(j==0?"":", "));
bbCat(stmt,tmp);
}
bbCat(stmt,"};");
codedump(stmt);
}
for(i=0;i<listlength(tsym->subnodes);i++) {
Symbol* efield = (Symbol*)listget(tsym->subnodes,i);
char tmp[1024];
ASSERT(efield->subclass == NC_FIELD);
#ifdef TESTALIGNMENT
snprintf(tmp,sizeof(tmp),"%lu",efield->typ.offset);
#else
snprintf(tmp,sizeof(tmp),"NC_COMPOUND_OFFSET(%s,%s)",
ctypename(tsym), cname(efield));
#endif
if(efield->typ.dimset.ndims > 0){
2010-06-03 21:24:43 +08:00
bbprintf0(stmt,"%sstat = nc_insert_array_compound(%s, %s, \"%s\", %s, %s, %d, %s_dims);",
indented(1),
groupncid(tsym->container),
typencid(tsym),
escapifyname(efield->name),
tmp,
typencid(efield->typ.basetype),
efield->typ.dimset.ndims,
cname(efield));
} else {
bbprintf0(stmt,"%sstat = nc_insert_compound(%s, %s, \"%s\", %s, %s);",
indented(1),
groupncid(tsym->container),
typencid(tsym),
escapifyname(efield->name),
tmp,
typencid(efield->typ.basetype));
}
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
}
codelined(1,"}");
break;
case NC_ARRAY:
/* ignore: this will be handled by def_var*/
break;
default: panic("genc_deftype: unexpected type subclass: %d",tsym->subclass);
}
}
#endif /*USE_NETCDF4*/
static void
genc_defineattr(Symbol* asym)
{
/* we need to capture vlen strings for dumping */
Bytebuffer* save = bbNew(); /* capture so we can dump
vlens first */
List* oldstate = NULL;
generator_getstate(c_generator,(void*)&oldstate);
listfree(oldstate);
generator_reset(c_generator,(void*)listnew());
generate_attrdata(asym,c_generator,(Writer)genc_write,save);
bbFree(save);
}
static void
genc_definevardata(Symbol* vsym)
{
Bytebuffer* save; /* capture so we can dump vlens first */
List* oldstate = NULL;
if(vsym->data == NULL) return;
save = bbNew();
generator_getstate(c_generator,(void*)&oldstate);
listfree(oldstate);
generator_reset(c_generator,(void*)listnew());
generate_vardata(vsym,c_generator,(Writer)genc_write,save);
bbFree(save);
}
static void
genc_write(Generator* generator, Symbol* sym, Bytebuffer* code,
int rank, size_t* start, size_t* count)
{
if(sym->objectclass == NC_ATT)
genc_writeattr(generator,sym,code,rank,start,count);
else if(sym->objectclass == NC_VAR)
genc_writevar(generator,sym,code,rank,start,count);
else
PANIC("illegal symbol for genc_write");
}
static void
genc_writevar(Generator* generator, Symbol* vsym, Bytebuffer* code,
int rank, size_t* start, size_t* count)
{
Symbol* basetype = vsym->typ.basetype;
2010-06-03 21:24:43 +08:00
nc_type typecode = basetype->typ.typecode;
List* vlendecls;
/* define a block to avoid name clashes*/
codeline("");
codelined(1,"{");
2010-06-03 21:24:43 +08:00
/* Dump any vlen decls first */
generator_getstate(generator,(void**)&vlendecls);
if(vlendecls != NULL && listlength(vlendecls) > 0) {
int i;
for(i=0;i<listlength(vlendecls);i++) {
Bytebuffer* decl = (Bytebuffer*)listget(vlendecls,i);
codelined(1,bbContents(decl));
bbFree(decl);
}
listfree(vlendecls);
generator_reset(generator,NULL);
}
2010-06-03 21:24:43 +08:00
if(rank == 0) {
codelined(1,"size_t count = 0;");
/* We make the data be an array so we do not need to
ampersand it later => we need an outer pair of braces
*/
commify(code); /* insert commas at proper places */
bbprintf0(stmt,"%sstatic %s %s_data[1] = {%s};\n",
indented(1),
ctypename(basetype),
cname(vsym),
bbContents(code));
codedump(stmt);
bbprintf0(stmt,"%sstat = nc_put_var1(%s, %s, &count, %s_data);\n",
indented(1),
groupncid(vsym->container),
varncid(vsym),
cname(vsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
codeflush();
} else { /* rank > 0 */
int i;
size_t length = 0;
if(typecode == NC_CHAR) {
length = bbLength(code);
/* generate data constant */
bbprintf(stmt,"%schar* %s_data = ",
indented(1),
cname(vsym),
(unsigned long)length);
codedump(stmt);
cquotestring(code,'"');
codedump(code);
codeline(" ;");
} else {
/* Compute total size */
length = 1;
for(i=0;i<rank;i++) length *= count[i];
/* generate data constant */
commify(code); /* insert commas at proper places */
bbprintf(stmt,"%s%s %s_data[%lu] = ",
indented(1),
ctypename(basetype),
cname(vsym),
(unsigned long)length);
codedump(stmt);
/* C requires an outer set of braces on datalist constants */
codepartial("{");
codedump(code);
codeline("} ;");
}
/* generate constants for startset, countset*/
bbprintf0(stmt,"%ssize_t %s_startset[%u] = {",
indented(1),
cname(vsym),
rank);
for(i=0;i<rank;i++) {
bbprintf(stmt,"%s%lu",(i>0?", ":""),start[i]);
}
codedump(stmt);
codeline("} ;");
bbprintf0(stmt,"%ssize_t %s_countset[%u] = {",
indented(1),
cname(vsym),
rank);
for(i=0;i<rank;i++) {
bbprintf(stmt,"%s%lu",(i>0?", ":""),count[i]);
}
codedump(stmt);
codeline("};");
bbprintf0(stmt,"%sstat = nc_put_vara(%s, %s, %s_startset, %s_countset, %s_data);\n",
indented(1),
groupncid(vsym->container), varncid(vsym),
cname(vsym),
cname(vsym),
cname(vsym));
codedump(stmt);
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
}
/* end defined block*/
codelined(1,"}\n");
codeflush();
}
static void
genc_writeattr(Generator* generator, Symbol* asym, Bytebuffer* code,
int rank, size_t* start, size_t* count)
{
Symbol* basetype = asym->typ.basetype;
int typecode = basetype->typ.typecode;
size_t len = asym->data->length; /* default assumption */
2010-06-03 21:24:43 +08:00
/* define a block to avoid name clashes*/
codeline("");
codelined(1,"{");
2010-06-03 21:24:43 +08:00
/* Handle NC_CHAR specially */
if(typecode == NC_CHAR) {
len = bbLength(code); /* presumably before quoting */
2011-11-14 12:20:19 +08:00
/* Revise length if length == 0 */
if(len == 0) len++;
cquotestring(code,'"');
2010-06-03 21:24:43 +08:00
} else {
/* All other cases */
/* Dump any vlen decls first */
List* vlendecls;
generator_getstate(generator,(void**)&vlendecls);
if(vlendecls != NULL && listlength(vlendecls) > 0) {
int i;
for(i=0;i<listlength(vlendecls);i++) {
Bytebuffer* decl = (Bytebuffer*)listget(vlendecls,i);
codelined(1,bbContents(decl));
bbFree(decl);
}
listfree(vlendecls);
generator_reset(generator,NULL);
}
2010-06-03 21:24:43 +08:00
commify(code);
bbprintf0(stmt,"%sstatic const %s %s_att[%ld] = ",
indented(1),
ctypename(basetype),
cname(asym),
asym->data->length
);
2010-06-03 21:24:43 +08:00
codedump(stmt);
codepartial("{");
codedump(code);
codepartial("}");
codeline(" ;");
bbClear(code);
}
/* Use the specialized put_att_XX routines if possible*/
switch (basetype->typ.typecode) {
case NC_BYTE:
case NC_SHORT:
case NC_INT:
case NC_FLOAT:
case NC_DOUBLE:
bbprintf0(stmt,"%sstat = nc_put_att_%s(%s, %s, \"%s\", %s, %lu, %s_att);\n",
indented(1),
ncstype(basetype->typ.typecode),
groupncid(asym->container),
(asym->att.var == NULL?"NC_GLOBAL"
:varncid(asym->att.var)),
escapifyname(asym->name),
typencid(basetype),
2010-06-03 21:24:43 +08:00
len,
cname(asym));
codedump(stmt);
break;
case NC_CHAR:
/* Include the string constant in-line */
bbprintf0(stmt,"%sstat = nc_put_att_%s(%s, %s, \"%s\", %lu, %s);\n",
indented(1),
ncstype(basetype->typ.typecode),
groupncid(asym->container),
(asym->att.var == NULL?"NC_GLOBAL"
:varncid(asym->att.var)),
escapifyname(asym->name),
len,
bbContents(code));
codedump(stmt);
break;
/* !usingclassic only (except NC_STRING) */
case NC_UBYTE:
case NC_USHORT:
case NC_UINT:
case NC_INT64:
case NC_UINT64:
2015-08-16 06:26:35 +08:00
if(usingclassic && k_flag <= 2) {
2010-06-03 21:24:43 +08:00
verror("Non-classic type: %s",nctypename(basetype->typ.typecode));
return;
}
bbprintf0(stmt,"%sstat = nc_put_att_%s(%s, %s, \"%s\", %s, %lu, %s_att);",
indented(1),
ncstype(basetype->typ.typecode),
groupncid(asym->container),
(asym->att.var == NULL?"NC_GLOBAL"
:varncid(asym->att.var)),
escapifyname(asym->name),
typencid(basetype),
2010-06-03 21:24:43 +08:00
len,
cname(asym));
codedump(stmt);
break;
2015-08-16 06:26:35 +08:00
#ifdef USE_NETCDF4
2010-06-03 21:24:43 +08:00
case NC_STRING:
if(usingclassic) {
verror("Non-classic type: %s",nctypename(basetype->typ.typecode));
return;
}
bbprintf0(stmt,"%sstat = nc_put_att_%s(%s, %s, \"%s\", %lu, %s_att);",
indented(1),
ncstype(basetype->typ.typecode),
groupncid(asym->container),
(asym->att.var == NULL?"NC_GLOBAL"
:varncid(asym->att.var)),
escapifyname(asym->name),
len,
cname(asym));
codedump(stmt);
break;
#endif
default: /* User defined type */
#ifndef USE_NETCDF4
verror("Non-classic type: %s",nctypename(basetype->typ.typecode));
2015-08-16 06:26:35 +08:00
#else /* !USE_NETCDF4 */
2010-06-03 21:24:43 +08:00
if(usingclassic && !isclassicprim(basetype->typ.typecode)) {
verror("Non-classic type: %s",nctypename(basetype->typ.typecode));
}
bbprintf0(stmt,"%sstat = nc_put_att(%s, %s, \"%s\", %s, %lu, %s_att);\n",
2010-06-03 21:24:43 +08:00
indented(1),
groupncid(asym->container),
(asym->att.var == NULL?"NC_GLOBAL"
:varncid(asym->att.var)),
escapifyname(asym->name),
typencid(basetype),
len,
cname(asym));
codedump(stmt);
#endif
break;
}
This PR adds EXPERIMENTAL support for accessing data in the cloud using a variant of the Zarr protocol and storage format. This enhancement is generically referred to as "NCZarr". The data model supported by NCZarr is netcdf-4 minus the user-defined types and the String type. In this sense it is similar to the CDF-5 data model. More detailed information about enabling and using NCZarr is described in the document NUG/nczarr.md and in a [Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in). WARNING: this code has had limited testing, so do use this version for production work. Also, performance improvements are ongoing. Note especially the following platform matrix of successful tests: Platform | Build System | S3 support ------------------------------------ Linux+gcc | Automake | yes Linux+gcc | CMake | yes Visual Studio | CMake | no Additionally, and as a consequence of the addition of NCZarr, major changes have been made to the Filter API. NOTE: NCZarr does not yet support filters, but these changes are enablers for that support in the future. Note that it is possible (probable?) that there will be some accidental reversions if the changes here did not correctly mimic the existing filter testing. In any case, previously filter ids and parameters were of type unsigned int. In order to support the more general zarr filter model, this was all converted to char*. The old HDF5-specific, unsigned int operations are still supported but they are wrappers around the new, char* based nc_filterx_XXX functions. This entailed at least the following changes: 1. Added the files libdispatch/dfilterx.c and include/ncfilter.h 2. Some filterx utilities have been moved to libdispatch/daux.c 3. A new entry, "filter_actions" was added to the NCDispatch table and the version bumped. 4. An overly complex set of structs was created to support funnelling all of the filterx operations thru a single dispatch "filter_actions" entry. 5. Move common code to from libhdf5 to libsrc4 so that it is accessible to nczarr. Changes directly related to Zarr: 1. Modified CMakeList.txt and configure.ac to support both C and C++ -- this is in support of S3 support via the awd-sdk libraries. 2. Define a size64_t type to support nczarr. 3. More reworking of libdispatch/dinfermodel.c to support zarr and to regularize the structure of the fragments section of a URL. Changes not directly related to Zarr: 1. Make client-side filter registration be conditional, with default off. 2. Hack include/nc4internal.h to make some flags added by Ed be unique: e.g. NC_CREAT, NC_INDEF, etc. 3. cleanup include/nchttp.h and libdispatch/dhttp.c. 4. Misc. changes to support compiling under Visual Studio including: * Better testing under windows for dirent.h and opendir and closedir. 5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags and to centralize error reporting. 6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them. 7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible. Changes Left TO-DO: 1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
codelined(1,"CHECK_ERR(stat);");
2010-06-03 21:24:43 +08:00
codelined(1,"}");
}
/* Compute the C name for a given symbol;
modified to use the fqn
*/
const char*
cname(Symbol* sym)
{
char* name;
assert (sym->fqn != NULL && sym->name != NULL);
/* Convert the fqn as its C name. */
if(sym->grp.is_root)
name = codify(sym->name);
else
name = codify(sym->fqn);
return name;
}
2010-06-03 21:24:43 +08:00
#endif /*ENABLE_C*/