Re: emacs rendering comparisson between emacs23 and emacs26.3

[Eli Zaretskii]

> From: Stefan Monnier <address@hidden>
> Cc: address@hidden,  Eli Zaretskii <address@hidden>,  address@hidden,
>   address@hidden,  address@hidden
> Date: Mon, 06 Apr 2020 11:25:43 -0400
> 
> >> 16s in one case a "10 seconds freeze": it sounds to me like you *do*
> >> reproduce the problem.
> > Not really.  In the 16s case redisplay kicks in and displays the buffer
> > 250 times.  In the 10s case redisplay stops after five times.
> 
> Yes, but the underlying problem remains: it takes 10-16s to do the work,
> so if the corresponding input events come in over a period shorter than
> 10-16s Emacs won't be able to keep up.

I think "the work" of a single scroll-up takes less (much less) than
10-16s.  But when several C-v keypresses accumulate (because the user
types them faster than Emacs can process them), their collective work
takes a long time to process.

> The text gets fontified by the scrolling routine because the size of the
> text depends on the fontification, so correct scrolling needs to fontify
> the text before it can compute the destination of the scroll.
> 
> That's where `fast-but-imprecise-scrolling` comes into play.

Right.