Re: Changes in revno 100600 on the emacs-23 branch

[YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp>]

>>>>> On Sat, 18 Jun 2011 19:40:44 +0300, Eli Zaretskii <eliz@gnu.org> said:

>> === modified file 'src/dispnew.c'
>> --- a/src/dispnew.c	2011-05-25 03:06:05 +0000
>> +++ b/src/dispnew.c	2011-06-18 00:37:38 +0000
>> @@ -5206,6 +5206,7 @@
>> /* Copy on the display.  */
>> if (r->current_y != r->desired_y)
>> {
>> +	    rif->clear_window_mouse_face (w);
>>  	    rif->scroll_run_hook (w, r);

> Why do you do this inside the loop?  Mouse highlight needs only be
> turned off once, right?  And no code inside the loop seems to turn it
> back on, right?

The first call clears dpyinfo->mouse_face_window to Qnil inside
clear_mouse_face, and thus the subsequent calls return early.  Of
course one could save the function call overhead by introducing a
local variable that differentiates the first call, but scroll_run_hook
involves graphics operations and is not light in the first place, and
I chose code similarity with the corresponding parts in try_window_id
and try_window_reusing_current_matrix.

> And btw, where's the code that will turn mouse highlight back on after
> this?

Just as in the case for try_window_id and
try_window_reusing_current_matrix, frame_up_to_date_hook will call
note_mouse_highlight.

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp