Re: platform-specific patches

[Ismael Luceno <ismael%iodev.co.uk@localhost>]

On 29/Dec/2024 21:42, Thomas Klausner wrote:
> On Sat, Dec 28, 2024 at 04:46:40PM -0700, Brook Milligan wrote:
> > 
> > > On Dec 28, 2024, at 12:53, Thomas Klausner <wiz%gatalith.at@localhost> wrote:
> > > 
> > > Instead of doing this, would it be possible to use the existing
> > > PATCH_* mechanism for this patchset?
> > > 
> > > You could easily set these variables conditionally inside an .ifdef.
> > > 
> > > Look at shells/bash for an example.
> > 
> > Sorry, I’m not quite sure what you are suggesting.
> > 
> > The PATCH_SITES and PATCHFILES mechanism allows to download
> > external patches into ${DISTDIR}.
> > 
> > If there were external patches, then different PATCHFILES for
> > different platforms would be easy to support as you note.  However,
> > the case I am trying to support is when there are no external patch
> > files, but different patches are required for different platforms.
> > 
> > The situation with gcc that motivated this is that the external
> > patchset for one platform (Darwin aarch64) conflicts with what
> > would be pkgsrc patches for another (Darwin x86_64).  Note that
> > the external patchset diverges from the upstream code base, is
> > really big, and hits lots of files, whereas the proposed pkgsrc
> > patch aligns with the upstream code base.  Presumably, there is a
> > good reason not to integrate the external patchset into pkgsrc,
> > but the resulting conflict prevents integrating minor upstream
> > changes into pkgsrc for Darwin x86_64.
> 
> The pkgsrc patches for Darwin x86_64 can easily be made a separate
> patchset and put on ftp.netbsd.org, and then only applied in the
> Darwin x86_64 case. That would solve this problem with minimal effort.

I wonder... How much effort would it be to make the patches universal?