On Sat, Aug 05, 2017 at 09:11PM, Aaron M. Ucko wrote:
> Builds of python-udatetime for a number of architectures failed, as
> excerpted below (modulo some varation in elapsed time) and detailed at
> Specifically, these failures occurred on armel, armhf, mips, mipsel,
> and the non-release architectures hppa, powerpc, ppc64, and sparc64.
> I'm not sure offhand why those specific architectures encountered
> these errors; I doubt the buildd's native timezone is relevant, since
> I see a successful mips64el build at mipsel-manda01 and a failed
> mipsel build at mipsel-manda03, which is presumably at the same data
> Could you please take a look?
It seems that an out-of-bounds read causes the above failures.
I have prepared a patch and will soon upload a new version.