"J. Lewis Muir" <jlmuir%imca-cat.org@localhost> writes: > On 1/27/15 4:26 AM, Benny Siegert wrote: >> It turns out the other pythons build because they have a compatibility >> stanza in their respective Makefiles. I added that to python34 and got >> it to build. > > Hi, Benny. > > I assume the compatibility stanza to which you refer is the one I > included in the patch in this email? > > http://mail-index.netbsd.org/pkgsrc-users/2015/01/07/msg020889.html > > If so, that's what Greg objects to. He would like there to be a general > fix for this, not a fix per package. Given that this is in the other python makefiles, it doesn't seem so horrific to add it to 34. But it should be CPPFLAGS, unless that doesn't work. The real problem is that there's an underlying bug in the python build process or the pkgsrc wrappers, because every program has to be passed -I for the header locations. Given that almost everything builds, it seems overwhelmingly likely that it's a python bug. (Seeing the volume of patches for python does not inspire confidence!) So adding CPPFLAGS seems to paper over that, but it's not really sound.
Attachment:
pgpWj0q9HHZqI.pgp
Description: PGP signature