Re: bug#80607: 31.0.50; notmuch-search with long query crashes Wayland

Subject: Re: bug#80607: 31.0.50; notmuch-search with long query crashes Wayland

Date: Sun, 15 Mar 2026 20:28:26 +0000

To: Eli Zaretskii

Cc: pipcet@protonmail.com, 80607@debbugs.gnu.org, notmuch@notmuchmail.org

From: Sean Whitton


Eli Zaretskii <eliz@gnu.org> [14/Mar  7:27pm +02] wrote:
>> From: Sean Whitton <spwhitton@spwhitton.name>
>> Cc: 80607@debbugs.gnu.org, notmuch@notmuchmail.org
>> Date: Sat, 14 Mar 2026 17:12:14 +0000
>>
>> Eli Zaretskii <eliz@gnu.org> [13/Mar  4:05pm +02] wrote:
>> > Please show a C-level backtrace to this effect.  If that's not
>> > possible "the easy way" (but I wonder why), step with GDB through
>> > redisplay_internal and find the line of code which causes the crash.
>>
>> I got the crash to happen under gdb and Emacs exits 1 because it loses
>> the display server connection but did not crash.
>
> That's expected.
>
>> David Bremner <david@tethera.net> [14/Mar  7:59am +09] wrote:
>> > One observation is that notmuch rather naively generates a buffer name
>> > from the query, so it might be possible to duplicate the problem without
>> > notmuch by generating a very long buffer name (possibly followed by some
>> > operations like pop-buffer-to-same-window).
>>
>> Good hint about setting window titles.  If I hack notmuch-search so that
>> it truncates the buffer name to 1000 characters then the problem is no
>> longer reproducible.  So that's what's going on.
>
> How long was the original string, FTR?

It's the whole search query plus a little more, so about 2000 characters.

-- 
Sean Whitton
_______________________________________________
notmuch mailing list -- notmuch@notmuchmail.org
To unsubscribe send an email to notmuch-leave@notmuchmail.org

Thread: