bug#21730: 25.0.50; Random errors in redisplay--pre-redisplay-functions

[Eli Zaretskii]

> From: Philipp Stephani <p.stephani2@gmail.com>
> Date: Wed, 14 Sep 2016 16:54:23 +0000
> 
>  Thanks for testing. I pushed the change, and I'm arking this bug as
>  done.
> 
> This change was reverted in 76ef52267cf887e3e1aa6d25b3b16dd0601dd459.

Too bad people revert changes that fixed a bug without making sure
they don't reintroduce the same bug again.  Now Emacs 25.1 will be
released with this bug.

> Given that replacing (bobp) with (= point 1) does solve this bug, the 
> documentation of pre-redisplay-functions
> must be incorrect, i.e. the current buffer is not the buffer of the window 
> passed as argument. I think the only
> way how this can happen is that a previous entry in pre-redisplay-functions 
> has changed the current buffer.
> Probably the implementation of redisplay--pre-redisplay-functions should be 
> changed from 
> (with-current-buffer (window-buffer win)
> (run-hook-with-args 'pre-redisplay-functions win))
> to
> (run-hook-wrapped 'pre-redisplay-functions
> (lambda (func) (with-current-buffer (window-buffer win)
> (funcall func win)
> nil))
> or so.
> 
> So we might try the following: Replace redisplay--pre-redisplay-functions 
> (and indeed, all hooks that
> document anything about the current buffer) as above. Then add (cl-assert (eq 
> (current-buffer)
> (window-buffer window)) to cursor-sensor--detect and remove code such as 
> (window-point window), because
> that must be equal to (point). WDYT? Or am I misunderstanding something?

Since the problem that caused the patch to be reverted was with
narrowing, why not simply use

  (= point (with-current-buffer (window-buffer window) (point-min)))

instead of

  (= point 1)

?

Anyway, my role in this bug was just to facilitate the debugging and
push a commit, I didn't really look at the issue deeper than what you
see in the discussion.