2.75.3: array-test fails on big-endian CPUs
After uploading GLib 2.75.3-1 to Debian experimental, the s390x architecture and the unofficial hppa, powerpc and ppc64 ports all have a test failure in /array/new/take-zero-terminated
at line 279. Of the architectures where a GLib 2.75.3 build has been attempted, these are exactly the architectures that are big-endian, suggesting that the CPU endianness is a factor.
This is from s390x, but the other failures look similar:
1/357 glib:glib+core / array-test RUNNING
>>> G_TEST_BUILDDIR=/<<PKGBUILDDIR>>/debian/build/deb/glib/tests MALLOC_CHECK_=2 G_DEBUG=gc-friendly G_ENABLE_DIAGNOSTIC=1 LD_LIBRARY_PATH=/<<PKGBUILDDIR>>/debian/build/deb/glib MALLOC_PERTURB_=167 G_TEST_SRCDIR=/<<PKGBUILDDIR>>/glib/tests /<<PKGBUILDDIR>>/debian/build/deb/glib/tests/array-test
▶ 1/357 /array/ref-count OK
▶ 1/357 /array/steal OK
▶ 1/357 /array/clear-func OK
▶ 1/357 /array/binary-search OK
▶ 1/357 /array/copy-sized OK
▶ 1/357 /array/overflow-append-vals OK
▶ 1/357 /array/overflow-set-size OK
▶ 1/357 /array/new/zero-terminated OK
▶ 1/357 /array/new/take OK
▶ 1/357 /array/new/take-zero-terminated - GLib:ERROR:../../../glib/tests/array-test.c:279:array_new_take_zero_terminated: assertion failed (garray->len == array_size): (254 == 255) FAIL
▶ 1/357 ERROR
1/357 glib:glib+core / array-test ERROR 0.01s killed by signal 6 SIGABRT
――――――――――――――――――――――――――――――――――――― ✀ ―――――――――――――――――――――――――――――――――――――
stderr:
**
GLib:ERROR:../../../glib/tests/array-test.c:279:array_new_take_zero_terminated: assertion failed (garray->len == array_size): (254 == 255)
(test program exited with status code -6)