[Python-checkins] r60892 - in python/branches/release25-maint: Lib/test/test_struct.py Misc/NEWS

["Martin v. Löwis"]

> Hm. I would have been happier with just disabling the test if
> sys.maxsize > some value. Having a regression test for this crasher
> has some value, even if the crasher is 32-bit specific.

I'm fine with adding it to the trunk. For a bug fix release, new
tests don't add that much value, IMO, in particular if they don't
get well-tested themselves before the release. IOW, if they are
clearly wrong as-committed (as the ones I removed), they are out.

FWIW, neither the 2.5 branch nor the trunk have sys.maxsize, so
using that as a predicate would not have worked. Using sys.maxint
(as someone proposed in the tracker) is bogus, as it would make
the test run incorrectly on Win64.

Regards,
Martin