Fedora Account System
Red Hat Associate
Red Hat Customer
Aforementioned tests fails on s390x while passes on other arches. Reproducible: Always Steps to Reproduce: 1. Ensure spec does not exclude s390x 2. fedpkg scratch-build --arches s390x --srpm Actual Results: "test-emoji-db" test fails Expected Results: All tests passes
Created attachment 2118264 [details] build.log
My guess is that the test is comparing native (big endian on s390x) data with a little endian baseline ...
The failing assert [1] is: > g_assert_cmpstrv (found, > ((const char *const[]) { > "\360\237\230\204\357\270\217", > "\360\237\230\201\357\270\217", > "\360\237\231\202\357\270\217", > NULL, > })); which spits out the actual and expected: > ERROR:../tests/test-emoji-db.c:21:test_emoji_db_match: assertion failed (found == ((const char *const[]) { "\360\237\230\204\357\270\217", "\360\237\230\201\357\270\217", "\360\237\231\202\357\270\217", NULL, })): first differing element at index 0: "\374\204\275\240\204\200\357\270\217" does not equal [1] https://gitlab.gnome.org/World/Phosh/stevia/-/blob/v0.51.0/tests/test-emoji-db.c?ref_type=tags#L21
This package has changed maintainer in Fedora. Reassigning to the new maintainer of this component.
Closing this. If anyone ever presents a valid use-case for running a mobile-focused OSK on their mainframe, we can revisit this.
I am fine with closing. Generally it's useful to assess bugs like this, because they might uncover more serious issues (we had a test failure that translated into a GCC 16 bug recently).
Thanks Dan, I generally agree with you. For example if someone gave me a repro case where a Phosh project doesn't build under a novel toolchain (i.e a very new GCC/Clang) on a supported platform, I'd consider that worth root-causing and fixing. In this case though, building these packages on s390x/ppc64le is quite wasteful in both time and resources, as it's extremely unlikely anyone will ever actually *use* these packages in a meaningful way on those platforms. If I were to chase down this particular ghost and it ended up uncovering a fundamental issue with GCC on s390x, that would indeed be ideal and I'd consider the effort worthwhile. But, I'm not really convinced that chances are very good that this will be the case :P