bug#64596: 30.0.50; On FIXME: in src/buffer.c:1481 (force-mode-line-update)

[Eli Zaretskii]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: yantar92@posteo.net,  64596@debbugs.gnu.org
> Date: Sun, 16 Jul 2023 14:53:03 -0400
> 
> >> Hmm... I'm not sure why you're saying
> >> 
> >>     frame's redisplay flag is what causes us to consider a frame in the
> >>     first place.
> >
> > Because fset_redisplay does this:
> >
> >   void
> >   fset_redisplay (struct frame *f)
> >   {
> >     redisplay_other_windows ();
> >     f->redisplay = true;
> >   }
> >
> > and redisplay_other_windows sets windows_or_buffers_changed, which
> > then causes consider_all_windows_p to be true upon the next redisplay
> > cycle:
> >
> >   consider_all_windows_p = (update_mode_lines
> >                         || windows_or_buffers_changed);
> 
> Yes, most calls to `[fbw]set_redisplay` will call
> `redisplay_other_windows` and once that function has been called, all
> the frames will be "considered", but that happens regardless of the
> `redisplay` bit of any particular frame.

If we always set these flags by calling these functions, then we will
end up doing more in redisplay than needed, because update_mode_lines
and windows_or_buffers_changed also disable certain optimizations.
When someone writes code that calls fset_redisplay, he or she are
likely to think that all this does is set the redisplay flag.  Which
is not true.  If every call to fset_redisplay indeed needs to disable
those optimizations, we should have more flags, and should probably
not call the function fset_redisplay, but something else.

> The `redisplay` bit is consulted later, once we loop over all the
> windows in all those frames, where we decide which of those windows are
> updated depending on the `redisplay` bits of the window, the window's
> buffer, and the window's frame.

That's not all of the uses of this flag.  Here's one example of other
uses:

          if (some_windows
              && !f->redisplay
              && !w->redisplay
              && !XBUFFER (w->contents)->text->redisplay)
            continue;

(This avoids updating the tool bar and the tab bar of the frame, and
there's a similar condition that avoids updating the frame's title.)

> Are you maybe confused by the name of the `FRAME_REDISPLAY_P` macro,
> which does *not* pay attention to the `redisplay` bit?

No, I don't think I'm confused about that.