Hi Ludovic,

ludo@gnu.org (Ludovic Courtès) writes:

[...]

>
> I’ve seen the failure once and it succeeded on a subsequent rebuild, so
> I came to the conclusion that this is a non-deterministic failure.  That
> would explain why the problem doesn’t show up on Hydra.
>
> Does it fail systematically for you?
>

Yes, it does! I can re-run the `guix system reconfigure ...' as
many times as I want and always get the same two failures. However, if I
keep the failed derivation and run the tests manually, I've found that
using 'strace' will cause more failures/errors; so it seems
non-deterministic; maybe a racing issue?

I've opened an issue on their GitHub tracker
(https://github.com/urwid/urwid/issues/230). Interestingly the same
tests run fine with Python 3, with or without using strace.

I've attached a patch that disables the problematic vterm tests, only
for the python2-urwid package.

Maxim