Re: Next steps for igc feature branch

[Eli Zaretskii]

> From: Sean Whitton <spwhitton@spwhitton.name>
> Cc: Pip Cet <pipcet@protonmail.com>,  Jonathon McKitrick <jcm@SDF.ORG>
> Date: Wed, 23 Apr 2025 09:38:50 +0800
> 
> Hello,
> 
> On Sat 19 Apr 2025 at 01:38pm GMT, Pip Cet via "Emacs development 
> discussions." wrote:
> 
> > I believe our priority at this point should still be (empirical)
> > stability across operating systems and in different use cases.  There
> > are some known crash scenarios (as the above-mentioned bug shows, the
> > same is true for the master branch), but we might have to live with them
> > for now (as we do on the master branch).
> >
> > As far as I know, there are no fundamental design questions left to be
> > settled or changes to be made.
> >
> > So I'm suggesting that we state somewhere that while there are still
> > known crash scenarios for the feature/igc branch, we expect them not to
> > occur in the wild (during ordinary interactive use).  So our response to
> > such crash reports should no longer be "we expected that to happen": if
> > a crash happens during ordinary use, clearly it's not just theoretical
> > and needs to be fixed.
> 
> After reading this, my thought is that unless the emacs-31 branch cut is
> imminent (which I do not believe it is), it would be appropriate to
> merge the igc branch to master.

Before we can seriously discuss landing this branch, we need to decide
what to do with the MPS library and how users will obtain and install
it on different systems.

Specifically, here are the issues I'm aware of (they were brought up
previously, but AFAIK never finalized):

 . Some platforms aren't supported by MPS.  We need to decide what to
   do about that.
 . Some of the supported platforms need patches to MPS that are not
   part of the upstream package, and in some cases weren't even
   submitted to MPS developers.  How do we expect users of those
   platforms to obtain and install MPS?  When we discussed this, some
   people said relying on distros to provide it might be unwise.
 . There were doubts about the future of MPS: whether it is still
   being actively maintained and supported.  Some people suggested
   that we fork and maintain the library ourselves, as part of Emacs.
 . The signal handling issue was resolved, but I'm not convinced the
   solution is the best one.  I think we should discuss this with the
   MPS developers, or at least decide whether we are okay with what we
   have now to stay with us.
 . There are still FIXMEs in igc.c.
 . We should have a clear idea which Emacs configurations, in terms of
   the toolkit and optional features, were sufficiently well tested.
 . Some people posted their TODO lists for the branch:

    https://lists.gnu.org/archive/html/emacs-devel/2025-01/msg00401.html
    https://lists.gnu.org/archive/html/emacs-devel/2025-01/msg00413.html

   Is any of that still relevant, and if so, should it be done before
   landing the branch?