Re-enabling these tests found that this bug (which is probably actually
multiple bugs) _does_ still exist, on big-endian systems and _possibly_
others.
pandas now (0.25.3+dfsg-7) warns the user when these are used on any
non-x86 system.
** Stata format
- All big-endian (s390x, hppa, ppc64):
tests/io/test_stata.py::TestStata::test_read_dta18, test_writer_117,
test_convert_strl_name_swap, test_unicode_dta_118, test_strl_latin1
fail. Suggests a problem involving Unicode strings.
- ppc64el: Succeeds on the buildd, but 34 tests (mostly but not only
involving nan/inf) fail in local qemu-ppc64el.
** HDF(5) format
- All big-endian:
tests/io/pytables/test_pytables.py::TestHDFStore::test_legacy_table_fixed_format_read_py2,
test_read_py2_hdf_file_in_py3, test_legacy_datetimetz_object fail.
Suggests an issue with reading older files, possibly the same pytables
bug (which doesn't appear to have a number) as
https://sources.debian.org/src/pytables/3.6.1-2/debian/patches/0005-Skip-index-backcompat-tests-on-bingendian.patch/
- s390x + ppc64 but not hppa: test_flush fails.
- All big-endian: test_encoding is skipped (by upstream).
(armhf may _crash_ - see #790925 - but is not known to give wrong results.)
** Clipboard
- tests/io/test_clipboard.py::test_raw_roundtrip failed in 0.25.3+dfsg-5
on s390x, but passed in -6 and -7 with no obvious fix in pandas.
Possibly a bug outside pandas that has now been fixed.
