Cursor doesn't go back to after close-paren

Cursor doesn't go back to after close-paren

[Juanma Barranquero]

emacs -Q --no-site-file

Type: ()

The blinking cursor remains over "(" until a key is pressed.

Is this related to the `sit-for' changes?

-- 
                    /L/e/k/t/u

Re: Cursor doesn't go back to after close-paren

[Chong Yidong]

"Juanma Barranquero" <lekktu@gmail.com> writes:

> emacs -Q --no-site-file
>
> Type: ()
>
> The blinking cursor remains over "(" until a key is pressed.
>
> Is this related to the `sit-for' changes?

I can't reproduce this.  Does it happen all the time for you?

Re: Cursor doesn't go back to after close-paren

[Eli Zaretskii]

> From: Chong Yidong <cyd@stupidchicken.com>
> Date: Thu, 03 Aug 2006 09:50:37 -0400
> Cc: Emacs Devel <emacs-devel@gnu.org>
> 
> "Juanma Barranquero" <lekktu@gmail.com> writes:
> 
> > emacs -Q --no-site-file
> >
> > Type: ()
> >
> > The blinking cursor remains over "(" until a key is pressed.
> >
> > Is this related to the `sit-for' changes?
> 
> I can't reproduce this.

I can.  It's probably specific to MS-Windows.  And it certainly looks
like it's related to sit-for, since moving the mouse to the tool bar
or the mode line (which generates a help-echo event) causes the cursor
to get back to after the right paren.

Damn that sit-for change!  Now will have 6 months of debugging such
subtle problems!

Re: Cursor doesn't go back to after close-paren

[Eli Zaretskii]

Can someone please describe in detail how the current sit-for is
supposed to work, in particular in the situation where it is called
from blink-matching-open?  Please describe in detail the OS-related
machinery, if any (interrupts, signals, timers, etc.) which are
involved; I'd like to compare that with what happens on MS-Windows, to
find out why the cursor doesn't return on Windows.

TIA

Re: Cursor doesn't go back to after close-paren

[Chong Yidong]

Eli Zaretskii <eliz@gnu.org> writes:

> Can someone please describe in detail how the current sit-for is
> supposed to work, in particular in the situation where it is called
> from blink-matching-open?  Please describe in detail the OS-related
> machinery, if any (interrupts, signals, timers, etc.) which are
> involved; I'd like to compare that with what happens on MS-Windows, to
> find out why the cursor doesn't return on Windows.

The sit-for mechanism no longer uses timers.  It basically relies on

  (read-event nil nil seconds)

The new SECONDS argument to `read-event', if non-nil, specifies a
number of seconds to wait for input.  If no input arrives within that
time, it stops waiting and returns nil.  That's probably what you
first want to make sure is working.

`read-event' works by passing a SECONDS argument to
read_filtered_event, which computes the desired end time and passes
that to read_char, which passes it to kbd_buffer_get_event, which
loops calling wait_reading_process_output.  During this loop,
kbd_buffer_get_event tells wait_reading_process_output to wait the
desired number of seconds (normally, when there is no SECONDS
argument, it tells wait_reading_process_output to wait indefinitely).
We break out of the kbd_buffer_get_event loop if the end time expires
(in which case we return Qnil), or if wait_reading_process_output
returns due to a valid input event.

The old sit-for, in contrast, simply called
wait_reading_process_output for the given number of seconds.

Re: Cursor doesn't go back to after close-paren

[Eli Zaretskii]

> Cc: lekktu@gmail.com,  emacs-devel@gnu.org
> From: Chong Yidong <cyd@stupidchicken.com>
> Date: Thu, 03 Aug 2006 16:14:43 -0400
> 
> `read-event' works by passing a SECONDS argument to
> read_filtered_event, which computes the desired end time and passes
> that to read_char, which passes it to kbd_buffer_get_event, which
> loops calling wait_reading_process_output.  During this loop,
> kbd_buffer_get_event tells wait_reading_process_output to wait the
> desired number of seconds (normally, when there is no SECONDS
> argument, it tells wait_reading_process_output to wait indefinitely).
> We break out of the kbd_buffer_get_event loop if the end time expires
> (in which case we return Qnil), or if wait_reading_process_output
> returns due to a valid input event.

Thanks, I think I fixed this (the problem was general, btw, not
specific to MS-Windows).  Patch below; I already committed it to CVS,
after testing on MS-Windows and on GNU/Linux (the latter only on a
TTY).  Please test more.

2006-08-05  Eli Zaretskii  <eliz@gnu.org>

	* keyboard.c (kbd_buffer_get_event): Return Qnil when current time
	is exactly equal to end_time, not only when it is past that.

Index: src/keyboard.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/keyboard.c,v
retrieving revision 1.863
diff -u -p -r1.863 keyboard.c
--- src/keyboard.c	5 Aug 2006 01:38:21 -0000	1.863
+++ src/keyboard.c	5 Aug 2006 11:58:32 -0000
@@ -3954,13 +3954,15 @@ kbd_buffer_get_event (kbp, used_mouse_me
 	{
 	  EMACS_TIME duration;
 	  EMACS_GET_TIME (duration);
-	  EMACS_SUB_TIME (duration, *end_time, duration);
-	  if (EMACS_TIME_NEG_P (duration))
-	    return Qnil;
+	  if (EMACS_TIME_GE (duration, *end_time))
+	    return Qnil;	/* finished waiting */
 	  else
-	    wait_reading_process_output (EMACS_SECS (duration),
-					 EMACS_USECS (duration), 
-					 -1, 1, Qnil, NULL, 0);
+	    {
+	      EMACS_SUB_TIME (duration, *end_time, duration);
+	      wait_reading_process_output (EMACS_SECS (duration),
+					   EMACS_USECS (duration),
+					   -1, 1, Qnil, NULL, 0);
+	    }
 	}
       else
 	wait_reading_process_output (0, 0, -1, 1, Qnil, NULL, 0);

Re: Cursor doesn't go back to after close-paren

[Michaël Cadilhac]

[-- Attachment #1.1: Type: text/plain, Size: 586 bytes --]

"Juanma Barranquero" <lekktu@gmail.com> writes:

> emacs -Q --no-site-file
>
> Type: ()
>
> The blinking cursor remains over "(" until a key is pressed.
>
> Is this related to the `sit-for' changes?

I can't reproduce it either.  What is your system ?

-- 
 |      Michaël `Micha' Cadilhac   |  Mieux vaut se taire                   |
 |         Epita/LRDE Promo 2007   |   Que de parler trop fort.             |
 | http://www.lrde.org/~cadilh_m   |           -- As de trèfle              |
 `--  -   JID: micha@amessage.be --'                                   -  --'

[-- Attachment #1.2: Type: application/pgp-signature, Size: 188 bytes --]

[-- Attachment #2: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-devel