Commit Graph

8066 Commits

Author SHA1 Message Date
Ward Fisher
379f51a68c Additional refactoring of the thredds dap4 test in support of #2077 2021-08-18 10:24:42 -06:00
Ward Fisher
306614a91a Renamed dap4_test thredds test file ins upport of https://github.com/Unidata/netcdf-c/issues/2077 2021-08-18 10:05:31 -06:00
Ward Fisher
87cac5995e path syntax tweak. 2021-08-17 14:16:04 -06:00
Ward Fisher
6516b19bbe Update the source url for windows installer downloads to reflect the move tot he new artifacts server. 2021-08-17 13:49:45 -06:00
Ward Fisher
c8f03b8910 Bump version numbers for the next development cycle. 2021-08-17 10:41:07 -06:00
Ward Fisher
7dec8f10cd Updated SO Version in line with https://www.gnu.org/software/libtool/manual/html_node/Updating-version-info.html 2021-08-17 10:39:00 -06:00
Ben Boeckel
78f9f5fbac cmake: extract the HDF5 version when finding HDF5 manually 2021-08-16 13:49:09 -04:00
Ward Fisher
e06d8b744f Merge branch 'patch-48' of https://github.com/gsjaardema/netcdf-c into v4.8.1-wellspring.wif 2021-08-16 10:37:52 -06:00
Ward Fisher
4b7c5a2bff Merge branch 'patch-49' of https://github.com/gsjaardema/netcdf-c into v4.8.1-wellspring.wif 2021-08-16 10:29:52 -06:00
Greg Sjaardema
f3041b3acf
Add a comment
Added comment indicating that we want integer division / truncation.
This should also trigger the tests to be run again; it looks like one of the tests just didn't run for some reason...
2021-08-16 10:06:45 -06:00
Ward Fisher
d1f2b970a0 Merge branch 'ejh_remote_benchmarks' of https://github.com/edwardhartnett/netcdf-c into v4.8.1-wellspring.wif 2021-08-13 16:13:47 -06:00
Ward Fisher
fc5f3d464b Merge branch 'ejh_dap_remote' of https://github.com/edwardhartnett/netcdf-c into v4.8.1-wellspring.wif 2021-08-13 16:08:17 -06:00
Ward Fisher
dc7b29556d Clean up orphaned file entry causing a failure in 'make distcheck' 2021-08-13 11:30:08 -06:00
Edward Hartnett
1b4322d88b turned off running of run_knmi_bm.sh benchmark with make check 2021-08-13 03:19:36 -06:00
Edward Hartnett
ac5d458c94 enable dap remote tests by default 2021-08-13 03:06:42 -06:00
Edward Hartnett
ff4ad14e8e disable dap remote tests by default 2021-08-13 03:03:26 -06:00
Ward Fisher
f676bb3575
Merge pull request #2072 from edwardhartnett/ejh_fix_2
made nc_perf not build and run benchmarks in parallel, turn off benchmark tst_gfs_data_1.c for HDF5 versions less than 1.10.2
2021-08-12 16:07:45 -06:00
Ward Fisher
b8a04fd8fa
Merge pull request #2071 from mathstuf/hdf5-library-variable-hole
hdf5: try and fill in HDF5_C_LIBRARY from HDF5_hdf5_LIBRARY if possible
2021-08-12 16:06:51 -06:00
Ward Fisher
1279681d36
Merge pull request #2066 from DennisHeimbigner/zarrfixes.dmh
Fix a number of bugs in the nczarr code.
2021-08-12 16:06:06 -06:00
Greg Sjaardema
28457cb570 Disable varible rename test
This PR adds the coordinate dimids to all variables; previously 1D
variables did not have it.  There is an issue with renaming variables
with the coordinate dimid which is now exposed in this test which
renames a 1D variable.

Disabling parts of this test until the variable renaming code can be
refactored.
2021-08-12 10:56:39 -06:00
Edward Hartnett
39de5eb6b5 added H5public.h to includes 2021-08-12 06:21:34 -06:00
Edward Hartnett
7353c8da1b turned of tst_gfs_data_1.c benchmark for HDF5 less than 1.10.2 2021-08-12 06:00:53 -06:00
Edward Hartnett
3aab4e284f made nc_perf not build and run benchmarks in parallel 2021-08-12 02:19:48 -06:00
Ben Boeckel
7727a68db4 hdf5: try and fill in HDF5_C_LIBRARY from HDF5_hdf5_LIBRARY if possible
A fairly vanilla build of 1.12.1 into a non-default directory ends up
with `HDF5_C_LIBRARY` set to `hdf5` which ends up failing all of the
`try_compile` checks because `-lhdf5` cannot be found.
2021-08-11 20:27:45 -04:00
Dennis Heimbigner
8be7d29335 Turn off bad test 2021-08-11 14:59:27 -06:00
Dennis Heimbigner
a58d243245 Fix library crash 2021-08-11 12:28:06 -06:00
Dennis Heimbigner
2a0fdb34d7 Fix RELEASE_NOTES conflicts 2021-08-10 15:02:38 -06:00
Dennis Heimbigner
aeeeb11e1e de-push 2021-08-10 14:53:52 -06:00
Dennis Heimbigner
3b13943e4d Add test cases 2021-08-10 14:15:24 -06:00
Ward Fisher
a193b444eb
Merge pull request #2065 from DennisHeimbigner/netpath.dmh
Support Windows network paths: \\svc\x\y...
2021-08-10 11:12:36 -06:00
Ward Fisher
0b5b484734
Merge pull request #2042 from mathstuf/nc3-support
dinfermodel: don't fail if NC_NETCDF4 is allowed, but NC3 is found
2021-08-10 11:11:22 -06:00
Ben Boeckel
f7a52e7fbd dinfermodel: don't fail if NC_NETCDF4 is allowed, but NC3 is found
In VTK, there are some files which require the NC3 implementation, but
no longer open under 4.8.0 (it worked under 4.7.4). The code checks to
make sure that certain formats were *not* requested when it is entirely
reasonable that support may be required for other files.

This partially reverts changes made in
59e04ae071 which is a massive commit which
adds Zarr support but doesn't mention why this specific change was made.
2021-08-10 07:40:29 -04:00
Dennis Heimbigner
42854d19de update release notes 2021-08-09 17:09:18 -06:00
Dennis Heimbigner
ec258bf314 Fix a number of bugs in the nczarr code.
re: Issues https://github.com/Unidata/netcdf-c/issues/2063, https://github.com/Unidata/netcdf-c/issues/2062, https://github.com/Unidata/netcdf-c/issues/2061, https://github.com/Unidata/netcdf-c/issues/2059

1. Support "fill_value: null" (https://github.com/Unidata/netcdf-c/issues/2063).
2. Handle the dtype case "|u1" (https://github.com/Unidata/netcdf-c/issues/2062).
3. When writing a pure Zarr format file, some nczarr attributes inadvertently crept in (https://github.com/Unidata/netcdf-c/issues/2061).
4. If there is no fill value, then the .zarray fill_value key should have the value null rather than left out (https://github.com/Unidata/netcdf-c/issues/2059).

Hat tip: Even Rouault
2021-08-09 17:05:02 -06:00
Ward Fisher
07b1464237
Merge pull request #2064 from edwardhartnett/ejh_mem1
Turn off tst_mem1.c test as part of benchmarks
2021-08-09 15:42:59 -06:00
Dennis Heimbigner
01487ecbf1 Update release notes 2021-08-09 15:37:42 -06:00
Dennis Heimbigner
de23473ac3 Support Windows network paths: \\svc\x\y...
re: Issue https:\\github.com\Unidata\netcdf-c\issues\2060

The path conversion code forgot to consider the case of
windows network paths of the form \\svc\x\y...

I have added support for it, but I can't really test it
since I do not have access to a network drive.
2021-08-09 15:34:23 -06:00
Edward Hartnett
78cc4743fa merged ejh_fix_warning 2021-08-09 13:46:10 -06:00
Edward Hartnett
b832ee3ad9 remove nc_perf/tst_mem1 from tests run 2021-08-09 13:43:17 -06:00
Edward Hartnett
3955376af6 remove nc_perf/tst_mem1 from tests run 2021-08-09 13:42:04 -06:00
Ward Fisher
a68d671686
Merge pull request #2058 from DennisHeimbigner/bmfix.dmh
Add missing files so --enable-benchmarks works again.
2021-08-09 11:54:41 -06:00
Dennis Heimbigner
e923d3965f Add missing files so --enable-benchmarks works again.
re: https://github.com/Unidata/netcdf-c/issues/2055

The nczarr_test benchmarks were missing some files and one was
out of date vis-a-vis compilation.
2021-08-07 21:19:40 -06:00
Ward Fisher
70862eb435
Merge pull request #2052 from e4t/build-fixes
check: Include testpathcvt.sh only if test is built
2021-08-05 10:14:16 -06:00
Greg Sjaardema
c3f20c88e9
Make variable easier to compress
Make the data in the variable `var1` easier to compress so that the compression tests will be more robust.  The zlib_ng library, for example, at level 1 does not compress a sequence of integers very well which resulted in the NetCDF-4 compressed file being larger than the uncompressed NetCDF-3 file.  With the change above, the variable contains `0,0,0,1,1,1,2,2,2,...,` which compresses at level 1 and the compression test is more robust.
2021-08-05 07:52:50 -06:00
Egbert Eich
6c4cb2c9f3 check: Include testpathcvt.sh only if test is built
This avoids the test case to fail if it isn't.

Signed-off-by: Egbert Eich <eich@suse.com>
2021-08-04 18:37:23 +02:00
Greg Sjaardema
b9d192d0c4
Only write the coord dimids if ndims >= 1
It looks like some vars have ndims==0 in which case the coord_dimids should not be written. Modify patch to catch those cases.
2021-08-04 09:49:48 -06:00
Greg Sjaardema
7b6f11c544
The coord dimids should be written for all variables
See discussion in #1279
2021-08-03 13:36:34 -06:00
Edward Hartnett
c77c4a9a40 fixing H5Linterate() API compatipility problem 2021-08-03 02:27:57 -06:00
Edward Hartnett
5d3e7a5147 fixed warning in tst_h_files4.c 2021-08-03 01:53:12 -06:00
Ward Fisher
84f0696e7d
Merge pull request #2036 from Unidata/gh1983.wif
Address optimization issue
2021-07-29 11:15:29 -06:00