Bug#1010202: libmatio: FTBFS against hdf5 1.12.0 currently in experimental - 3 failed tests

Gilles Filippini pini at debian.org
Tue May 17 21:26:45 BST 2022


Hi Sébastien,

Gilles Filippini a écrit le 02/05/2022 à 11:04 :
> Hi,
> 
> Gilles Filippini a écrit le 29/04/2022 à 14:10 :
>> Hi Sébastien,
>>
>> Sébastien Villemot a écrit le 29/04/2022 à 11:45 :
>>> Hi Gilles,
>>>
>>> Le mardi 26 avril 2022 à 10:38 +0200, Gilles Filippini a écrit :
>>>> Source: libmatio
>>>> Version: 1.5.23-1
>>>> Severity: important
>>>> Tags: ftbfs
>>>>
>>>> While rebuilding libmatio 1.5.23 against hdf5 1.12.0, 3 testcases 
>>>> failed with:
>>> […]
>>>
>>> I had an exchange with upstream, who says that libmatio 1.5.23 should
>>> pass all tests against hdf5 1.12.2.
>>>
>>> Is there any reason why you’re sticking to hdf5 1.12.0 instead of the
>>> latest stable release 1.12.2?
>>
>> No, no reason. I wanted to transition 1.12.0 before switching to 
>> 1.12.2, but the other way should be fine as well. I'll give it a try.
> 
> I've just tested a rebuild against HDF5 1.12.2, and the same 3 tests 
> fail mosly the same way:
> 
> 2272. mat73_compressed_read_le.at:475: testing Read directory ...
> ./mat73_compressed_read_le.at:478: cp $srcdir/results/dir_le.out expout
>           $builddir/test_mat directory 
> $srcdir/datasets/matio_test_cases_compressed_hdf_le.mat
> --- /dev/null   2022-04-25 15:38:39.000000000 +0000
> +++ /<<PKGBUILDDIR>>/test/testsuite.dir/at-groups/2272/stderr 2022-05-02 
> 08:55:13.515220334 +0000
> @@ -0,0 +1,40 @@
> +-E- test_mat: HDF5 error #000 in H5Fopen()
> +      file : ../../../src/H5F.c:620
> +      major: File accessibility
> +      minor: Unable to open file
> +-E- test_mat: HDF5 error #001 in H5VL_file_open()
> +      file : ../../../src/H5VLcallback.c:3501
> +      major: Virtual Object Layer
> +      minor: Iteration failed
> +-E- test_mat: HDF5 error #002 in H5PL__path_table_iterate()
> +      file : ../../../src/H5PLpath.c:578
> +      major: Plugin for dynamically loaded library
> +      minor: Iteration failed
> +-E- test_mat: HDF5 error #003 in H5PL__path_table_iterate_process_path()
> +      file : ../../../src/H5PLpath.c:620
> +      major: Plugin for dynamically loaded library
> +      minor: Can't open directory or file
> +-E- test_mat: HDF5 error #004 in H5VL__file_open()
> +      file : ../../../src/H5VLcallback.c:3351
> +      major: Virtual Object Layer
> +      minor: Can't open object
> +-E- test_mat: HDF5 error #005 in H5VL__native_file_open()
> +      file : ../../../src/H5VLnative_file.c:97
> +      major: File accessibility
> +      minor: Unable to open file
> +-E- test_mat: HDF5 error #006 in H5F_open()
> +      file : ../../../src/H5Fint.c:1898
> +      major: File accessibility
> +      minor: Unable to lock file
> +-E- test_mat: HDF5 error #007 in H5FD_lock()
> +      file : ../../../src/H5FD.c:1625
> +      major: Virtual File Layer
> +      minor: Unable to lock file
> +-E- test_mat: HDF5 error #008 in H5FD__sec2_lock()
> +      file : ../../../src/H5FDsec2.c:1002
> +      major: Virtual File Layer
> +      minor: Unable to lock file
> +-E- test_mat: HDF5 error #000 in H5Fclose()
> +      file : ../../../src/H5F.c:707
> +      major: Invalid arguments to routine
> +      minor: Inappropriate type

Running each failed test alone doesn't fail. Hence I've just tried adding:

override_dh_auto_test:
	dh_auto_test --no-parallel

and tada! no more failure.

Best,

_g.



More information about the debian-science-maintainers mailing list