Re: [Emacs-diffs] master 5c9304e: Disable some display optimizations when frames need redisplay

[Eli Zaretskii <eliz@gnu.org>]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: emacs-devel@gnu.org
> Date: Mon, 05 Oct 2015 11:50:03 -0400
> 
> > Bonus points for adding something that describes where these variables
> > and the corresponding 'redisplay' flags are checked, because the
> > FOR_EACH_FRAME loop in redisplay_internal doesn't look at the
> > 'redisplay' flags, so a large part of the logic you describe above
> > isn't (easily) visible at that level.
> 
> I added a note about the fact that the frame's redisplay flag is not
> sufficient to exclude a frame.

Thanks.

> Fundamentally, the only redisplay flags needed are the window ones.
> The frame and buffer's redisplay flags could be dispensed with, at the
> cost of making [bf]set_redisplay more expensive since they'd have to
> loop through all the windows of that frame/buffer.

Yes, I agree.  But is that justified?  At least for the frame's flag,
a test for each window of the frame's window tree has the same cost as
setting the window-specific flag in fset_redisplay.  However, the test
is made only once per redisplay cycle, whereas the setting could be
done several times, depending on what Lisp is being run.  Right?