Re: Emacs [scratch/igc 6682d0e6c96] crash on Linux 6.6.41 KDE 6.0.5, wayland

[Pip Cet]

"Eli Zaretskii" <eliz@gnu.org> writes:

>> Date: Tue, 3 Sep 2024 22:58:04 +0800
>> From: execvy@gmail.com
>>
>> What file are you visiting when this happened?
>> A 
>> /home/exec/Projects/git.savannah.gnu.org/git/emacs-build/scratch_igc/share/emacs/31.0.50/lisp/files.el
>>  file.
>>
>> Another information: my KDE plasma setting has a keyboard config.
>> I config keyboard long press repeat rate as: 100 repeats/s
>> If I decrease 100 repeats/s to 10 repeats/s, then I feel no gc pause on 
>> scratch/igc.
>>
>> But, in master branch, even I set 100 repeats/s for long press key config. I 
>> feel no pause.
>
> And I don't feel any pauses with the igc branch, either.

I'm not sure key repeat is a particularly good test of GC
responsiveness; it appears to be heavily influenced by non-GC factors,
and when holding C-n, redisplay stops at some point and only catches up
when the key is released.

If I set the keyboard repeat rate to 255 using 'xset', which appears to
be the maximum supported, an MPS Emacs running my production session
(some 200 buffers) ends up exhibiting this behavior when holding C-n, so
only the first, non-repeated command is displayed properly.  A
newly-launched MPS Emacs does fine with xdisp.c in fundamental mode, but
in C mode it ends up scrolling for a bit, presumably until font-lock
does its work, then not scrolling until 'n' is released again.

It would be nice to change this behavior to force some redisplay even
when Emacs can't keep up with keypresses, but I don't think this is an
MPS issue at this point.

Is it true that we only redisplay once all key presses have been
handled?

Pip