Re: [PATCH] Ignore pending_signals when checking for quits.

[Philipp Stephani <p.stephani2@gmail.com>]

Am Do., 3. Jan. 2019 um 15:03 Uhr schrieb Eli Zaretskii <eliz@gnu.org>:
>
> > From: Philipp Stephani <p.stephani2@gmail.com>
> > Date: Wed,  2 Jan 2019 22:22:18 +0100
> > Cc: Philipp Stephani <phst@google.com>
> >
> >  static bool
> >  module_should_quit (emacs_env *env)
> >  {
> >    MODULE_FUNCTION_BEGIN_NO_CATCH (false);
> > -  return (! NILP (Vquit_flag) && NILP (Vinhibit_quit)) || pending_signals;
> > +  return QUITP;
> >  }
>
> Bother.  I see your point regarding the return value when just
> pending_signals is set, but disregarding pending_signals doesn't sound
> TRT to me, either.  It means various Emacs features based on input
> detection won't work while the module code runs, even if the module
> tries to be nice and does call module_should_quit.  For example,
> while-no-input and atimers won't work, and Emacs will generally be
> much less responsive to user input.
>
> So maybe we should indeed return true only if QUITP says so, but we
> should also call process_pending_signals from module_should_quit, when
> pending_signals is non-zero?

Wouldn't that mean that Emacs could do something (e.g. process
events)? That wouldn't match the naming and intention of should_quit:
By its name, it should only query information and not change any
state.