Re: Scrolling commands and skipping redisplay, was: Re: emacs rendering comparisson between emacs23 and emacs26.3

[Dmitry Gutov]

On 12.04.2020 10:10, Eli Zaretskii wrote:
> I cannot imagine why would we want to exempt only the scrolling
> commands from the input_was_pending trick.

Scrolling is certainly special at least in some way:

- Implementation-wise, not all commands simulate redisplay during their 
operation. Given that the heavy lifting has probably already been done, 
maybe skipping redisplay doesn't help save a lot of work anyway? At 
least without extra tricks like jit-lock-defer-time which speed up both 
simulated and "real" redisplay, making skipping redisplay less essential 
anyway.

- As we talked about "precision", the user needs to see the buffer 
contents to see where they are scrolling to, and where to stop. If the 
number of screens you need to scroll is >1, you can likely never decide 
on the necessary number of scrolls in advance. This is not the case for 
many other Emacs commands. E.g., if I use find-file, I usually don't 
need to see the prompt, and I can start typing the desired file name in 
advance.

> And if we remove it, what to do ab out the scenario
> described by Stefan in the large comment to that variable?

In a way, the comment supports my point:

  ...all the hard work of jit-locking is "wasted"...

The current approach helps avoid this most of the time, but it errs on 
the side of skipping. I suggest it might be better to err on the side of 
not skipping, for a certain set of commands at least.