bug#17588: 24.3.91; mouse-face glitchy

[Eli Zaretskii <eliz@gnu.org>]

> From: Christopher Schmidt <ch@ristopher.com>
> Date: Mon, 26 May 2014 04:35:18 -0400 (EDT)
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> > How was your Emacs built?  (The information collected by
> > 'report-emacs-bug' was not present in your report.)
> 
> No explicit configuration options were used.
> 
> I used a recent build of the emacs-24 branch for this report.  AFAICT
> bug reports via M-x report-emacs-bug RET in emacs-24 do not expose
> configuration features beyond what is in system-configuration-options.

Sorry, I didn't see the telltale "Configured with:" heading, and
assumed that you elided that part.  That heading is present in the
branch version as well.

> I do not have access to any other machine with a window system at the
> moment.
> 
> This one:
> 
>     $ grep "model name" /proc/cpuinfo | uniq
>     model name	: Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz

This is even slightly better than mine (Core i7-2600 CPU @ 3.40).  And
on top of that, your build is optimized, so should be roughly twice
faster than what I'm running.

IOW, this is hardly about CPU power, at least not as far as Emacs's
application thread is concerned.  It is also probably X specific.

Can you see how much of your CPU resources are busy when you run the
recipe?  When I run it here, I see about 12% to 25% of a single
execution unit occupied by Emacs.  Do you see something different, and
do you see some other process (e.g., the X server) pegging the CPU,
perhaps?

Another thought is to see whether the function that paints and clears
the mouse highlight is called when it should be.  You could add some
printf to note_mouse_highlight, where it calls show_mouse_face and
clear_mouse_face, and see if the corresponding print-outs are visible
when they should be.  E.g., if you move the mouse pointer from one
line of text with mouse face to another, you should see a call that
clears mouse face on the previous line and shows it on the current
one.

Thanks.