bug#39081: 26.3; Emacs hangs if external SCIM input method is killed

[Vladimir Nikishkin]

Does Emacs have any deliberate support for SCIM. (Why should it?)

Almost every other application simply ignores the crash. If input is requested again, SCIM daemon seems to get started on demand.

Eli Zaretskii <eliz@gnu.org> 於 2020年1月14日 週二 23:52 寫道：

> From: Vladimir Nikishkin <lockywolf@gmail.com>
> Date: Tue, 14 Jan 2020 11:54:15 +0800
> Cc: 39081@debbugs.gnu.org
>
> This is the "crashed" trace. Note the GTK message in between.

Thanks.  This seems to be an X error.

And now I'm actually asking myself how reasonable is it to expect
Emacs to continue working when its IM service crashes.  Why is this a
use case we should support? does it happen a lot in practice.

(I admit I know very little about the architecture of SCIM support in
Emacs, in terms of what OS services we use to communicate with it, and
thus I cannot reason about the difficulty in surviving SCIM crashes.
Does someone know?)