[Python-modules-team] Bug#860656: python-biplist: FTBFS on i386: dh_auto_test: pybuild --test --test-nose -i python{version} -p 2.7 returned exit code 13

[Christoph Biedl]

Lucas Nussbaum wrote...

> During a rebuild of all packages in stretch (in a stretch chroot, not a
> sid chroot), your package failed to build on i386.

[ reformatted ]
> > AssertionError: 
| '[4294967295L, 4294967294L, 4294967296L, 4294967293L, 4294967297L, 8589934590L, 2147483647L]' != 
| '[4294967295L, 4294967294L, 4294967296L, 4294967293L, 4294967297L, 8589934590L, 2147483647]'
                                                                                            ^

No fix for this issue yet, but I'd like to share the result of my
investigation.

This happens on more 32bit archs, I've seen this on armhf and powerpc as
well.

Building the stretch/sid version fails on both stretch and jessie.
Building the jessie version passes on both stretch and jessie.
This suggests the underlying problem is actually in this module, not in
plistlib.py as it might seem.

The expected string differs among architectures: On amd64, it's

| exp:[4294967295, 4294967294, 4294967296, 4294967293, 4294967297, 8589934590, 2147483647]

while on i386 it's rather

| [4294967295L, 4294967294L, 4294967296L, 4294967293L, 4294967297L, 8589934590L, 2147483647L]

since the intial value (4294967295) is a long and does not loose that
attribute even if it drops below 2^31 (2147483647L). On the contrary,
the readPlist routine appearently uses the minimum required data type,
resulting in the test failure as seen above.

Where this actually happens and the proper way to deal with this is left
for another day.

    Christoph
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/python-modules-team/attachments/20170506/4b65ab61/attachment-0001.sig>