Test failures on s390x

[opoplawski]

I'm looking to update the Fedora g2c library package to 2.1.0 but I'm getting the following test failures on s390x:

      Start 25: tst_metacmp
24/42 Test #25: tst_metacmp ......................***Failed    0.00 sec
Testing g2c_metacmp
      Start 26: tst_inq
      Start 27: tst_degrib2_int
26/42 Test #26: tst_inq ..........................***Failed    0.00 sec
Testing g2c inquiry functions.
Testing g2c_inq()/g2c_inq_msg()/g2c_inq_prod() calls...
	testing open of data/gdaswave.t00z.wcoast.0p16.f000.grib2 without an index file...
      Start 28: tst_degrib2_fast
28/42 Test #28: tst_degrib2_fast .................***Failed    0.00 sec
Testing g2c degrib2 function.
Testing g2c_degrib2() on file data/gdaswave.t00z.wcoast.0p16.f000.grib2...
      Start 31: tst_index
30/42 Test #30: tst_degrib2_ftp_all ..............   Passed    0.00 sec
      Start 32: tst_files
31/42 Test #31: tst_index ........................***Failed    0.00 sec
Testing g2c index functions.
Testing g2c_open_index() on file data/gdaswave.t00z.wcoast.0p16.f000.grib2...
      Start 33: tst_g2_testfile
32/42 Test #32: tst_files ........................***Failed    0.00 sec
Testing g2c file functions.
Testing g2c_open() on file data/gdaswave.t00z.wcoast.0p16.f000.grib2 (twice)...ok!
Testing g2c_find_msg2 on file data/gdaswave.t00z.wcoast.0p16.f000.grib2...
      Start 34: tst_aec
33/42 Test #33: tst_g2_testfile ..................***Failed    0.00 sec
Testing with file created in NCEPLIBS-g2.
testing opening of the file...
      Start 37: tst_large_png
37/42 Test #37: tst_large_png ....................***Failed    0.00 sec
Testing dec_png() on large test file...
      Start 42: run_compare_tests.sh
42/42 Test #42: run_compare_tests.sh .............***Failed    0.00 sec
*** Running g2c_convert test
g2c_compare 2.1.0 comparing data/gdaswave.t00z.wcoast.0p16.f000.grib2 and data/gdaswave.t00z.wcoast.0p16.f000.grib2.

This is likely due to the different endianness of s390x. There is not a lot of output from the tests so hard to know what is going on. Can we get more output?

[opoplawski]

I'm also seeing similar failures on 32-bit i686

[edwardhartnett]

Howdy Orion!

Firstly we are up to version 4.1.0, so you might as well be working with the latest version.

But I don't know how to test on machine of different endianness. Does GitHub offer any platform that would help?

[opoplawski]

Actually, looks like you are at version 2.1.0 - I think we both got confused there. Github does not appear to offer any s390x hosts for CI. I can build and run anything on a Fedora s390x machine - so let me know what would be helpful.