re: issue https://github.com/Unidata/netcdf-c/issues/1251
Assume that you have the URL to a remote dataset
which is a normal netcdf-3 or netcdf-4 file.
This PR allows the netcdf-c to read that dataset's
contents as a netcdf file using HTTP byte ranges
if the remote server supports byte-range access.
Originally, this PR was set up to access Amazon S3 objects,
but it can also access other remote datasets such as those
provided by a Thredds server via the HTTPServer access protocol.
It may also work for other kinds of servers.
Note that this is not intended as a true production
capability because, as is known, this kind of access to
can be quite slow. In addition, the byte-range IO drivers
do not currently do any sort of optimization or caching.
An additional goal here is to gain some experience with
the Amazon S3 REST protocol.
This architecture and its use documented in
the file docs/byterange.dox.
There are currently two test cases:
1. nc_test/tst_s3raw.c - this does a simple open, check format, close cycle
for a remote netcdf-3 file and a remote netcdf-4 file.
2. nc_test/test_s3raw.sh - this uses ncdump to investigate some remote
datasets.
This PR also incorporates significantly changed model inference code
(see the superceded PR https://github.com/Unidata/netcdf-c/pull/1259).
1. It centralizes the code that infers the dispatcher.
2. It adds support for byte-range URLs
Other changes:
1. NC_HDF5_finalize was not being properly called by nc_finalize().
2. Fix minor bug in ncgen3.l
3. fix memory leak in nc4info.c
4. add code to walk the .daprc triples and to replace protocol=
fragment tag with a more general mode= tag.
Final Note:
Th inference code is still way too complicated. We need to move
to the validfile() model used by netcdf Java, where each
dispatcher is asked if it can process the file. This decentralizes
the inference code. This will be done after all the major new
dispatchers (PIO, Zarr, etc) have been implemented.
Primary fixes to get -ansi to work.
1. Convert all '//' C++ style comments to /*...*/ or to use #if 0...#endif
2. It turns out that when -ansi is specified, then a number of
functions no longer are defined in the header -- but they are still
in the .so file.<br>
The big example is strdup(). So, added code to include/ncconfig.h to define
externs for those missing functions that occur in more than one place.
These are enabled if !_WIN32 && __STDC__ == 1 (__STDC__ is supposed to
be the equivalent compile time flag to -ansi). Note that this requires
config.h (which references ncconfig.h) to be included in files where it is
currently not included. Single uses will be only in the file that uses them.
3. Added mmap test for the MAP_ANONYMOUS flag to configure.ac. Apparently
this is not always defined with -ansi.
4. fix some large integer constants in nc_test4/tst_atts3.c and nc_test4/tst_filterparser.c
to avoid compiler complaints.
5. fix a double constant in nc_test4/tst_filterparser.c to avoid compiler complaints.
[Note I suspect #4 and #5 will be a problem on big-endian machines, but we have no way to test]
Misc. Changes:
1. convert more instances of _MSC_VER to _WIN32.
2. added some debugging code to include/nctestserver.h
3. added comment about libdispatch/drc.c always being compiled.
4. modify parser generation in ncgen to remove unneeded files.
re: issue https://github.com/Unidata/netcdf-c/issues/1233
Changes:
1. remove exit that was there for testing.
2. the program tst_open_mem must be netcdf-4 only.
3. fix some diff problems
- Change dataset name for tst_inmemory4_create to tst_inmemory4
- Modify tst_inmemory.c to reorder the variables (somewhat major rewrite)
Minor Unrelated Fixes:
1. fix comment problem in nc_provenance.h
2. Fix memory leak in tst_open_mem.c
3. fix ncdump/bindata.c to properly compile if netcdf4 is disabled.
4. minor changes to ncgen.l
re: pull request https://github.com/Unidata/netcdf-c/pull/1219
This pr should go in after or at the same time as 1219.
It updates nc_test/tst_inmemory to add a test to cover
the case fixed in 1219. It also disables a couple of tests
that no longer are relevant.
By increasing the DIM1 size, the created in-memory file will be
larger than the default memory allocation size (4096), so the code
will have to reallocate the memory which results in more code coverage.
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.
ret: https://github.com/Unidata/netcdf-c/issues/1162
The test nc_test/run_diskless2.sh fails
when LARGE_FILE_TESTS is enabled.
Since the goal of the test was to test out
diskless+persist on a reasonably large file,
I fixed by just limiting the file size to
1000000000L bytes.
https://github.com/Unidata/netcdf-c/issues/1168https://github.com/Unidata/netcdf-c/issues/1163https://github.com/Unidata/netcdf-c/issues/1162
This PR partially fixes memory leaks in the netcdf-c library,
in the ncdump utility, and in some test cases.
The netcdf-c library now runs memory clean with the assumption
that the --disable-utilities option is used. The primary remaining
problem is ncgen. Once that is fixed, I believe the netcdf-c library
will run memory clean with no limitations.
Notes
-----------
1. Memory checking was performed using gcc -fsanitize=address.
Valgrind-based testing has yet to be performed.
2. The pnetcdf, hdf4, and examples code has not been tested.
Misc. Non-leak changes
1. Make tst_diskless2 only run when netcdf4 is enabled (issue 1162)
2. Fix CmakeLists.txt to turn off logging if ENABLE_NETCDF_4 is OFF
3. Isolated all my debug scripts into a single top-level directory
called debug
4. Fix some USE_NETCDF4 dependencies in nc_test and nc_test4 Makefile.am
re: https://github.com/Unidata/netcdf-c/issues/1154
Inadvertently, the behavior of NC_DISKLESS with nc_create() was
changed in release 4.6.1. Previously, the NC_WRITE flag needed
to be explicitly used with NC_DISKLESS in order to cause the
created file to be persisted to disk.
Additional analyis indicated that the current NC_DISKLESS
implementation was seriously flawed.
This PR attempts to clean up and regularize the situation with
respect to NC_DISKLESS control. One important aspect of diskless
operation is that there are two different notions of write.
1. The file is read-write vs read-only when using the netcdf API.
2. The file is persisted or not to disk at nc_close().
Previously, these two were conflated. The rules now are
as follows.
1. NC_DISKLESS + NC_WRITE means that the file is read/write using the netcdf API
2. NC_DISKLESS + NC_PERSIST means that the file is persisted to a disk file at nc_close.
3. NC_DISKLESS + NC_PERSIST + NC_WRITE means both 1 and 2.
The NC_PERSIST flag is new and takes over the obsolete NC_MPIPOSIX flag.
NC_MPIPOSIX is still defined, but is now an alias for the NC_MPIIO flag.
It is also now the case that for netcdf-4, NC_DISKLESS is independent
of NC_INMEMORY and in fact it is an error to specify both flags
simultaneously.
Finally, the MMAP code was fixed to use NC_PERSIST as well.
Also marked MMAP as deprecated.
Also added a test case to test various combinations of NC_DISKLESS,
NC_PERSIST, and NC_WRITE.
This PR affects a number of files and especially test cases
that used NC_DISKLESS.
Misc. Unrelated fixes
1. fixed some warnings in ncdump/dumplib.c
In turn that caused some problems elsewhere because some files
did not include netcdf.h preceding netcdf_mem.h. Specifically,
building under visual studio fails.
So fix those cases that I can find. May need to
add more later.
re: github issue https://github.com/Unidata/netcdf-c/issues/1111
One of the less common use cases for the in-memory feature is
apparently failing with HDF5-1.10.x. The fix is complicated and
requires significant changes to libhdf5/nc4memcb.c. The current
setup is detailed in the file docs/inmeminternal.dox.
Additionally, it was discovered that the program
nc_test/tst_inmemory.c, which is invoked by
nc_test/run_inmemory.sh, actually was failing because of the
above problem. But the failure is not detected since the script
does not return non-zero value.
Other Changes:
1. Fix nc_test_tst_inmemory to return errors correctly.
2. Make ncdap_tests/findtestserver.c and dap4_tests/findtestserver4.c
be generated from ncdap_test/findtestserver.c.in.
3. Make LOG() print output to stderr instead of stdout to
avoid contaminating e.g. ncdump output.
4. Modify the handling of NC_INMEMORY and NC_DISKLESS flags
to properly handle that NC_DISKLESS => NC_INMEMORY. This
affects a number of code pieces, especially memio.c.
stored in the _NCProperties attribute to allow two things:
1. capture of additional library dependencies (over and above
hdf5)
2. Recognition of non-netcdf libraries that create netcdf-4 format
files.
To this end, the _NCProperties format has been extended to be
and arbitrary set of key=value pairs separated by commas.
This new format has version = 2, and uses commas as the pair separator.
Thus the general form is:
_NCProperties = "version=2,key1=value,key2=value2..." ;
This new version is accompanied by a new ./configure option of the form
--with-ncproperties="key1=value1,key2=value2..."
that specifies pairs to add to the _NCProperties attribute for all
files created with that netcdf library.
At this point, what is missing is some programmatic way to
specify either all the pairs or additional pairs
to the _NCProperties attribute. Not sure of the best way
to do this.
Builders using non-netcdf libraries can specify
whatever they want in the key value pairs (as long
as the version=2 is specified first).
By convention, the primary library is expected to be the
the first pair after the leading version=2 pair, but this
is convention only and is neither required nor enforced.
Related changes:
1. Fixed the tests that check _NCProperties to properly operate with version=2.
2. When reading a version 1 _NCProperties attribute, convert it to look
like a version 2 attribute.
2. Added some version 2 tests to ncdump/tst_fileinfo.c and
ncdump/tst_fileinfo.sh
Misc Changes:
1. Fix minor problem in ncdap_test/testurl.sh where a parameter to
buildurl needed to be quoted.
2. Minor fix to ncgen to swap switches -H and -h to be consistent
with other utilities.
3. Document the -M flag in nccopy usage() and the nccopy man page.
4. Modify a test case to use the nccopy -M flag.
The fix includes the following changes.
1. Checking and using the default file format at file create time is now
done only when the create mode (argument cmode) does not include any
format related flags, i.e. NC_64BIT_OFFSET, NC_64BIT_DATA,
NC_CLASSIC_MODEL, and NC_NETCDF4.
2. Adjustment of cmode based on the default format is now done in
NC_create() only. The idea is to adjust cmode before entering the
dispatcher's file create subroutine.
3. Any adjustment of cmode is removed from all I/O dispatchers, i.e.
NC4_create(), NC3_create(), and NCP_create().
4. Checking for illegal cmode has been done in check_create_mode() called
in NC_create(). This commit removes the redundant checking from
NCP_create().
5. Remove PnetCDF tests in nc_test/tst_names.c, so it can focus on testing
all classic formats and netCDF4 formats.
Two new test programs are added. They can be used to test netCDF with and
without this commit.
1. nc_test/tst_default_format.c
2. nc_test/tst_default_format_pnetcdf.c (use when PnetCDF is enabled).
re: github issue https://github.com/Unidata/netcdf-fortran/issues/82
This was originally discovered in the Fortran tests, but is
a problem in the C library.
The problem only occurred when using HDF5-1.10.x. The reason it
failed is that starting with 1.10, the hid_t type was changed
from 32 bits to 64 bits.
The function libsrc4/nc4memcb.c#NC4_image_init was using type int (doh!)
to return the hdf fileid instead of hid_t type. This, of course,
caused the id to be truncated and in turn later use of the id
caused hdf5 to fail.
Fix is trivial: replace int with hid_t. This also requires a related
change in nc4mem.c.
Also added the test case derived from the original Fortran code.
You would think I would learn...
corresponding HDF5 operations.
re: github issue https://github.com/Unidata/netcdf-c/issues/908
also in reference to https://github.com/pydata/xarray/issues/2004
The netcdf-c library has implemented the nc_get_vars and nc_put_vars
operations as element at a time. This has resulted in very slow
operation.
This pr attempts to improve the situation for netcdf-4/hdf5 files
by using the slab operations provided by the hdf5 library. The new
implementation passes the get/put vars stride information down to
the hdf5 slab operations.
The result appears to improve performance significantly. Some simple
tests on large 2-D arrays shows speedups in excess of 150.
Misc. other changes:
1. fix bug in ncgen/semantics.c; using a list's allocated length
instead of actual length.
2. Added a temporary hook in the netcdf library plus a performance
test case (tst_varsperf.c) to estimate the speedup. After users
have had some experience with this, I will remove it, probably
after the 4.7 release.
Fix https://github.com/Unidata/netcdf-c/issues/962
1. remove the --disable-diskless option since it is no
longer needed. Similarly for CMakeLists.txt.
2. Fixed nc4files.c where BAIL and return were mixed
leading to situation where cleanup code was not
being invoked. This probably occurs elsewhere,
but I did not find any specifically.
The file docs/indexing.dox tries to provide design
information for the refactoring.
The primary change is to replace all walking of linked
lists with the use of the NCindex data structure.
Ncindex is a combination of a hash table (for name-based
lookup) and a vector (for walking the elements in the index).
Additionally, global vectors are added to NC_HDF5_FILE_INFO_T
to support direct mapping of an e.g. dimid to the NC_DIM_INFO_T
object. These global vectors exist for dimensions, types, and groups
because they have globally unique id numbers.
WARNING:
1. since libsrc4 and libsrchdf4 share code, there are also
changes in libsrchdf4.
2. Any outstanding pull requests that change libsrc4 or libhdf4
are likely to cause conflicts with this code.
3. The original reason for doing this was for performance improvements,
but as noted elsewhere, this may not be significant because
the meta-data read performance apparently is being dominated
by the hdf5 library because we do bulk meta-data reading rather
than lazy reading.
and https://github.com/Unidata/netcdf-c/issues/708
Expand the NC_INMEMORY capabilities to support writing and accessing
the final modified memory.
Three new functions have been added:
nc_open_memio, nc_create_mem, and nc_close_memio.
The following new capabilities were added.
1. nc_open_memio() allows the NC_WRITE mode flag
so a chunk of memory can be passed in and be modified
2. nc_create_mem() allows the NC_INMEMORY flag to be set
to cause the created file to be kept in memory.
3. nc_close_mem() allows the final in-memory contents to be
retrieved at the time the file is closed.
4. A special flag, NC_MEMIO_LOCK, is provided to ensure that
the provided memory will not be freed or reallocated.
Note the following.
1. If nc_open_memio() is called with NC_WRITE, and NC_MEMIO_LOCK is not set,
then the netcdf-c library will take control of the incoming memory.
This means that the original memory block should not be freed
but the block returned by nc_close_mem() must be freed.
2. If nc_open_memio() is called with NC_WRITE, and NC_MEMIO_LOCK is set,
then modifications to the original memory may fail if the space available
is insufficient.
Documentation is provided in the file docs/inmemory.md.
A test case is provided: nc_test/tst_inmemory.c driven by
nc_test/run_inmemory.sh
WARNING: changes were made to the dispatch table for
the close entry. From int (*close)(int) to int (*close)(int,void*).
In C, `char`, `signed char`, and `unsigned char` are three separate,
distinct types, so just because `char` happens to be signed does not
mean it is interchangeable with `signed char`.
