Joerg Sonnenberger <joerg%bec.de@localhost> writes: > Am Wed, Nov 16, 2022 at 12:18:46PM +0000 schrieb Greg Troxel: >> Module Name: pkgsrc >> Committed By: gdt >> Date: Wed Nov 16 12:18:46 UTC 2022 >> >> Modified Files: >> pkgsrc/print/py-reportlab: distinfo >> Added Files: >> pkgsrc/print/py-reportlab/patches: patch-setup.py >> >> Log Message: >> print/py-reportlab: Remediate incorrect searching for ft2build.h >> >> setup.py grovels paths in the filesystem to find ft2build.h, >> apparently not consulting the provided CPPFLAGS. Add /usr/X11R7, just >> before /usr/local, to fix building on NetBSD 9 with base X11. Thanks >> to wiz@ for the hint. > > What about only looking into .buildlink instead? I didn't think of that. That would be a better fix, but it would involve a much greater rototill of setup.py, which would be a pkgsrc-specific, rather than staying in the spirit of upstream. I was trying to rock the boat minimally, as I expect that a larger-scale change would have fallout. I'm not sure how to deal with buildlink for libs, though, as the normal approach is for the program's build system to use the eventual paths, and the wrappers to redirect to buildlink. Alternatively, we could just rip this out entirely, since we have arranged via bl3 for freetype to be present, and there's no real point in provoking an early failure vs a later build failure.
Attachment:
signature.asc
Description: PGP signature