The process of updating version informaiton (in src/nagg.h) and in various
document files (e.g. README, RELEASE.txt) were done by hand. Automize most
parts of this process. Post reminder of files, e.g., doc/nagg.docx, that
require manual changes.
Also show post-release tasks.
Tested: koala (svn diff the effects and built successfully.)
src/nagg_get_granules.c:
Fixed bug: get_granules() did not check for maximum number of granules limit,
thus ended up in seg. fault when the limit was exceeded. Fixed by adding
the checking code and error messages.
src/nagg.h:
raise the limit of number of granules managed (NAGG_Granule_info_max)
to 10,000.
Tested: koala.
Bug fixed: was not filtering/selecting the latest version of granules.
Fixed by creating a new granprod_compar which compares granules ID and Product ID
only, but not granules version. Fixed another error by changing granule_compar
to compare Granule ID and Product ID in ascending order but version information values
in descending order. This corrected sorting order and the granprod_compar together
filter/select the latest version of every granule-product pairs.
Tested: koala.
test/testnagg.sh
test/testfiles
Add test and testfiles for problem with duplicate granules having identical ids but different granule versions.
doc/RELEASE.txt
Add description of problem to known problems section of RELEASE.txt.
Make check will fail with this test added.
Tested on jam and koala.
test/Makefile.in:
test/testnagg_more.sh:
test/Makefile.am:
test/nagg_functions.sh:
Bug fix: somehow testnagg.sh and testnagg_more.sh run in parallel (e.g.,
when "gmake -j 4" is used.) They both used the same style of temporary files
in the same test/ directory, resulting in overwritting each other.
Changed Makefile.am to force testnagg.sh and testnagg_more.sh to run in
sequencing. Also append pid ($$) to the temporary files which should avoid
the two test scripts from interferring each oth...
Move the function definitions from testnagg.sh to nagg_functions.sh so that
the function definitions can be reused by other Nagg test scripts to be
created.
Tested: jam, koala (both srcdir and inplace.)
nagg_get_granules.c
Add H5Gclose call to get_geo_granules() for a subgroup that was not closed, resulting in failure to open files with granule numbers above ~1600.
Added check of status returned from get_group_granules() in same function.
nagg_write_granules.c
Commented out check of number of granules written in end_write() function. For packaged REDRO-GCRIO the number written doesn't match the expected number, causing nagg to fail and stop running. It seems the error is incor...
Removed the limit of maximum number of input files.
The parse_option does not copy the input file arguments from argv to inputfiles
any more. It just assign the remaining part of argv to inputfiles.
Keep the limit for future use in case input file arg can be a dir.
Tested: koala and Jam.
Added code to transform the time-dependent Creation Date information in output
file names to generic format so that expected files can match the output
generated. The Creation date is not relevant to what we try to verify here.
testnagg.sh:
Added TRANSFORM to change Creation time in filenames to some generic format.
test/testfiles/satms_tatms2.txt
/testfiles/redroS.txt
test/testfiles/satms8.txt
test/testfiles/redro3S.txt
test/testfiles/redro1.txt
test/testfiles/redro3.txt
test/testfile...
nagg_get_granules.c
Eliminated segmentation faults for --debug and large numbers of files.
Still doesn't work for over 1600 GEO files, but works for --debug with 7200 REDRO files and 2000+ SVM* + GMODO files.
Tested on koala.
test/testnagg.sh:
Previous tests only check exit code. Improve testing by adding two new
features to testnagg.sh that will verify the expoected output for some
tests:
CHECK_OUTPUT:
run the tool test and expect succeed (exit code 0) with expected output
CHECK_FAIL_OUTPUT:
run the tool test and expect failure (exit code 1) with expected output
Current expected output verification can handle fixed output only. Next
improvement is to verify time-dependent (e.g., th...
nagg_write_granules.c
Added several missing H5Dclose statements.
nagg_get_granules.c
Added code to support --debug option to print all granules in input files and in the geolocation files referenced by the input files.
Added missing H5Fclose and H5Gclose statements that caused H5Fopen to fail because too many files were open.
Tested amani, jam & koala, gcc & icc.
Added the --debug option which, when given, prints all the granules in
all input files. This includes fill granules plus geolocation granules
if asked for. It only reads and prints granules in the input files and
does not do any further processing.
Tested: koala
Added long options (--xxx) feature.
Added the recongization of --help, --number, --product, --geolocation.
Added the parsing of the --debug optoin. (not doing anything different yet).
Added debug_arg variable to show --debug is used or not.
Tested: Koala and Jam. (only test for --help is tested).
Add to configuure to check for the getopt.h header file which is used by
getopt_long function.
Note: The previous commit was for the addion to confiugre for the checking of the getopt_long functioion.
Tested: koala using gcc, icc and pgcc.
-This line, and those below, will be ignored--
M configure
M configure.ac
M src/nagg_config.h.in
test/testnagg.sh
Add test for datasets with fixed dimensions.
Verifies fix for JPSS-41.
test/testiles
Add 1 SVI02 and 2 SVI03 files with fixed dimension datasets.
Tested jam & koala, gcc and icc.
Replace function copy_dataset_except_data() with function create_resized_duplicate_dataset.
Files were found with fixed dimensions causing H5Sset_extent_simple to fail when attempting
to increase the size of a fixed size dataset. The new function creates a dataset in the
output file duplicating the characteristics of the dataset in the input file with the
appropriate size for the number of granules in the new aggregation.
Remove local function reset_dataspace_extent which ...
