Re: Unifying code for drawing on a cairo context

[Po Lu]

"Pfrommer, Julius" <julius.pfrommer@web.de> writes:

> Cairo is not unmaintained. There were a couple hundred commits from
> various developers since the beginning of this year.
>
> https://github.com/freedesktop/cairo/graphs/contributors?from=2021-04-02&to=2022-04-29&type=c

The Cairo developers haven't been able to put out a stable release with
fixes to major bugs in over 3 years, which is interfering with the
development of many large programs and toolkits.  Most of those commits
you allude to are either deleting unbuildable code, or fixing the tests,
and are useless because so far the maintainer(s) capable of making a
release are all missing, so the code doesn't get to anyone.

> Case in point. For GTK and PGTK, developers need different setups (using
> X vs. using Wayland) and the similar code-paths for drawing on cairo_t
> started to diverge and continues to diverge.

And that divergence has caused no problems whatsoever.

> It appears that increasing the amount of code-sharing between platforms
> is a goal for Emacs overall.

That makes sense in the general case, but as the platform-specific
maintainer for both platforms, I disagree this is a direction we want to
work towards in the Cairo code.  In contrast, using the GDK selections
API is a similar refactoring task that would be much more worth spending
your time on, and one that I hope someone will complete soon.