[Python-modules-team] Bug#544291: makes packages on (at least) alpha FTBFS

Andreas Barth aba at not.so.argh.org
Sun Aug 30 14:10:01 UTC 2009


Hi,

* Kumar Appaiah (a.kumar at alumni.iitm.ac.in) [090830 16:03]:
> Thanks for notifying us of this. I am well aware of what is causing
> the problem, and know how to fix it. However, I wanted the opinion of
> the Debian Python team for a more elegant fix, before going ahead and
> fixing this problem. Please give me a couple of days, within which I
> can get this issue sorted out.

For debian purposes, I think it's really enough to just assume
endian.h is always there.

For upstream purposes, I don't think it is. I proposed
#include <stddef.h>
#if defined(NPY_HAVE_ENDIAN_H) || defined(__GLIBC__)

for that - but well, I don't know enough C to know if that's really a
good solution.

Unfortunatly this comes now at a bad time as this blocks the
transition I'm trying to get rid of since ~4 weeks now, so I would
propose to use the unelegant method of doing "#if 1" for now (or just
setting NPY_HAVE_ENDIAN_H unconditionally at the top of that file),
and refine it to something that works on non-gnu-platforms as well and
get it upstream incorporated afterwards.



Cheers,
Andi





More information about the Python-modules-team mailing list