Re: CVS commit: pkgsrc/graphics/cairo

[nia <nia%NetBSD.org@localhost>]

On Wed, Nov 22, 2023 at 02:27:17PM +0000, nia wrote:
> On Wed, Nov 22, 2023 at 09:06:53AM -0500, Greg Troxel wrote:
> > nia <nia%NetBSD.org@localhost> writes:
> > 
> > > On Wed, Nov 22, 2023 at 06:47:14AM -0500, Greg Troxel wrote:
> > >> Also, I don't follow "just to be safe".  If cairo documents that it
> > >> needs gnu99, then that's how it is.   If it doesn't, and it needs it,
> > >> then it's buggy and someone(tm) should file a bug and include the URL.
> > >
> > > USE_LANGUAGES=    c99 previously would imply c99-with-extensions-enabled,
> > > i.e. gnu99.
> > >
> > > Certain platforms (e.g. SunOS) are very restrictive when extensions
> > > are disabled, carefully checking which functions are used in the
> > > software. I do not currently have acess to these platforms. Therefore,
> > > I am uncomfortable unconditionally setting strict standards mode when
> > > software expects (as cairo does) to be compiled with default settings
> > > (usually gnu99).
> > 
> > Does cairo document that it needs gnu99?
> 
> Of course not. It needs a c99 compiler and by uses the default settings
> of that compiler, which in GCC and Clang means extensions will be enabled
> - i.e. gnu99.

I agree with the idea that it's a bug that cairo uses c99 features without
specifying -std=x99, but I also don't think they'd be very receptive
to the that it's a bug (the likely response is "get a newer compiler"),
and if I opened bug reports for every legacy-platform build issue I fix
in pkgsrc, I'd be able to do nothing with my life except try to win over
software authors!