On Sun, Aug 30, 2009 at 03:44:07PM +0200, Andreas Barth wrote: > tag 544291 + patch > thanks > > * Debian Bug Tracking System (ow...@bugs.debian.org) [090830 14:17]: > > [...] > > The obvious solution is to replace > #ifdef NPY_HAVE_ENDIAN_H > by > #if 1 > in numpy/core/include/numpy/npy_endian.h > (and drop NPY_HAVE_ENDIAN_H again from CFLAGS)
If you want to, I can upload it earlier than that as a (co-)maintainer. I just intended to find out if a more elegant method existed. > The not-so-primitive solution (but possible to push it upstream) could > be (however, I'm not insisting this is the correct one - I'm not so > firm on "what can assume cross-platform"): > --- ./numpy/core/include/numpy/npy_endian.h~ 2009-04-05 10:09:20.000000000 > +0200 > +++ ./numpy/core/include/numpy/npy_endian.h 2009-08-30 17:42:02.000000000 > +0200 > @@ -5,8 +5,9 @@ > * NPY_BYTE_ORDER is set to the same value as BYTE_ORDER set by glibc in > * endian.h > */ > +#include <stddef.h> > > -#ifdef NPY_HAVE_ENDIAN_H > +#if defined(NPY_HAVE_ENDIAN_H) || defined(__GLIBC__) > /* Use endian.h if available */ > #include <endian.h> > #define NPY_BYTE_ORDER __BYTE_ORDER > > (and also drop NPY_HAVE_ENDIAN_H again from CFLAGS) > > I'm intending to NMU with the first one next weekend, unless I hear > different. (The first one because it's obvious that it will do, and > because I won't push it upstream anyways.) I've sent this patch upstream. Thanks. Kumar -- Kumar Appaiah
signature.asc
Description: Digital signature
