Re: Calling Lisp from undo.c's record_* functions

[Eli Zaretskii <eliz@gnu.org>]

> From: Stefan Monnier <monnier@IRO.UMontreal.CA>
> Cc: phillip.lord@russet.org.uk, emacs-devel@gnu.org
> Date: Tue, 17 Nov 2015 21:52:49 -0500
> 
> > How about the following idea: we don't start the timer from
> > run_undoable_change; instead, we set a flag there that will be checked
> > by the command loop when it finishes execution of a command, and the
> > call to start the timer will be made then?  We don't really need to
> > attempt to start the timer for each and every change of every buffer,
> > do we?
> 
> That would work similarly to an idle timer, so yes, that would be OK
> (tho an idle timer would be simpler).  It suffers from the same problem
> as the idle timer: if there's no user interaction but there's continued
> process output, we'll fail to re-run the undo-auto-timer after the
> first N seconds of idleness.  Nothing too serious (it's still better
> than what we have now anyway), tho.

So we could start the timer not in the command loop, but somewhere in
the loop that waits for process output, perhaps.

But if an idle timer is fine, I think Phillip already has everyuthing
sorted out on a branch.