gnus 5.8.4 - display bug (same as in 5.8.3)

gnus 5.8.4 - display bug (same as in 5.8.3)

[Dmitry Yaitskov]

Hi,

This is IIRC the same (or very similar) bug which prevented me from
ever using 5.8.3: Enter a newsgroup (the *Summary* buffer), w/out
displaying any message yet. Now, go to some message in the list which
is in the lower part of the screen. Press Enter to show the message.
The window is split, the message is displayed in the lower part, but
the *Summary* buffer in the upper part of the screen is *not* scrolled
to keep the current message visible (it was in the lower part of the
screen, which is now taken my the Article buffer). The cursor parks on
some other message which is visible (w/out scrolling) when the
*Summary* window split. On my system this happens all the time (i.e.
100% reproducible).

-- 
Cheers,
-Dima.

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Lars Magne Ingebrigtsen]

Dmitry Yaitskov <dimas@home.com> writes:

> This is IIRC the same (or very similar) bug which prevented me from
> ever using 5.8.3: Enter a newsgroup (the *Summary* buffer), w/out
> displaying any message yet. Now, go to some message in the list which
> is in the lower part of the screen. Press Enter to show the message.
> The window is split, the message is displayed in the lower part, but
> the *Summary* buffer in the upper part of the screen is *not* scrolled
> to keep the current message visible (it was in the lower part of the
> screen, which is now taken my the Article buffer). The cursor parks on
> some other message which is visible (w/out scrolling) when the
> *Summary* window split. On my system this happens all the time (i.e.
> 100% reproducible).

I'm utterly unable to reproduce this with XEmacs 21.2 (beta16).  Has
stuff changed in this area in XEmacs?  (That is, `window-start' and
friends.) 

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Colin Rafferty]

Lars Magne Ingebrigtsen writes:
> Dmitry Yaitskov <dimas@home.com> writes:

>> This is IIRC the same (or very similar) bug which prevented me from
>> ever using 5.8.3: Enter a newsgroup (the *Summary* buffer), w/out
>> displaying any message yet. Now, go to some message in the list which
>> is in the lower part of the screen. Press Enter to show the message.
>> The window is split, the message is displayed in the lower part, but
>> the *Summary* buffer in the upper part of the screen is *not* scrolled
>> to keep the current message visible (it was in the lower part of the
>> screen, which is now taken my the Article buffer). The cursor parks on
>> some other message which is visible (w/out scrolling) when the
>> *Summary* window split. On my system this happens all the time (i.e.
>> 100% reproducible).

> I'm utterly unable to reproduce this with XEmacs 21.2 (beta16).  Has
> stuff changed in this area in XEmacs?  (That is, `window-start' and
> friends.) 

I have the same problem in 21.2 (beta32), the same vesion Dmitry is
using.

There were some definite changes in redisplay.  I think that it
doesn't redisplay until later than it used to.  We also do pixel-based
scrolling, which screwed up shrink-window-if-larger-than-buffer' (a
fix to which I am testing).

Do you do anything with window-pixel-edges?  I had to remove that.

-- 
Colin

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Dmitry Yaitskov]

Lars Magne Ingebrigtsen <larsi@gnus.org> wrote:

> Dmitry Yaitskov <dimas@home.com> writes:
> 
> > This is IIRC the same (or very similar) bug which prevented me from
> > ever using 5.8.3: Enter a newsgroup (the *Summary* buffer), w/out
> > displaying any message yet. Now, go to some message in the list which
> > is in the lower part of the screen. Press Enter to show the message.
> > The window is split, the message is displayed in the lower part, but
> > the *Summary* buffer in the upper part of the screen is *not* scrolled
> > to keep the current message visible (it was in the lower part of the
> > screen, which is now taken my the Article buffer). The cursor parks on
> > some other message which is visible (w/out scrolling) when the
> > *Summary* window split. On my system this happens all the time (i.e.
> > 100% reproducible).
> 
> I'm utterly unable to reproduce this with XEmacs 21.2 (beta16).  Has
> stuff changed in this area in XEmacs?  (That is, `window-start' and
> friends.) 

I have no idea, honestly. I'm using 21.2 beta 32, and I've seen the
bug (with gnus 5.8.3) in an earlier beta (beta 27 perhaps). I can try
to help you debug it if you tell me where and what to look for. I'd
very much like this resolved, as the bug is annoying enough (at least
with the way I mostly use gnus) to prevent me from using versions
having it.

-- 
Cheers,
-Dima.

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Björn Torkelsson]

Dmitry Yaitskov <dimas@home.com> writes:

> I have no idea, honestly. I'm using 21.2 beta 32, and I've seen the
> bug (with gnus 5.8.3) in an earlier beta (beta 27 perhaps). I can try
> to help you debug it if you tell me where and what to look for. I'd
> very much like this resolved, as the bug is annoying enough (at least
> with the way I mostly use gnus) to prevent me from using versions
> having it.

Try to set gnus-xmas-force-redisplay to nil. 

See the other thread about "Incorrect point after window splitting.

/torkel

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Lars Magne Ingebrigtsen]

Colin Rafferty <colin@xemacs.org> writes:

> There were some definite changes in redisplay.  I think that it
> doesn't redisplay until later than it used to.  We also do pixel-based
> scrolling, which screwed up shrink-window-if-larger-than-buffer' (a
> fix to which I am testing).
> 
> Do you do anything with window-pixel-edges?  I had to remove that.

I don't use that -- just `window-height', I think.  Perhaps that is
being computed later than it used to be?

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Shenghuo ZHU]

[-- Attachment #1: Type: text/plain, Size: 1036 bytes --]

>>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

LMI> Colin Rafferty <colin@xemacs.org> writes:
>> There were some definite changes in redisplay.  I think that it
>> doesn't redisplay until later than it used to.  We also do pixel-based
>> scrolling, which screwed up shrink-window-if-larger-than-buffer' (a
>> fix to which I am testing).
>> 
>> Do you do anything with window-pixel-edges?  I had to remove that.

LMI> I don't use that -- just `window-height', I think.  Perhaps that is
LMI> being computed later than it used to be?

As I remember, just before 5.8.3 was released, it was OK.  The
difference is that switch-to-buffer was changed to set-buffer in
gnus-configure-windows. I don't know the reason, but the attached
patch works for me (in XEmacs 21.2.32). This patch has not been
applied to the CVS repository. If it really works, could anyone tell
me the reason?

-- 
Shenghuo

2000-04-20 22:24:04  Shenghuo ZHU  <zsh@cs.rochester.edu>

	* gnus-win.el (gnus-configure-windows): Revert to switch-to-buffer.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: switch-to-buffer.diff --]
[-- Type: text/x-patch, Size: 750 bytes --]

Index: gnus-win.el
===================================================================
RCS file: /usr/local/cvsroot/gnus/lisp/gnus-win.el,v
retrieving revision 5.8
diff -u -r5.8 gnus-win.el
--- gnus-win.el	2000/01/09 15:38:16	5.8
+++ gnus-win.el	2000/04/21 02:23:08
@@ -446,11 +446,11 @@
 		    (gnus-delete-windows-in-gnusey-frames))
 		;; Just remove some windows.
 		(gnus-remove-some-windows)
-		(set-buffer nntp-server-buffer))
+		(switch-to-buffer nntp-server-buffer))
 	    (select-frame frame)))
 
 	(let (gnus-window-frame-focus)
-	  (set-buffer nntp-server-buffer)
+	  (switch-to-buffer nntp-server-buffer)
 	  (gnus-configure-frame split)
 	  (when gnus-window-frame-focus
 	    (select-frame (window-frame gnus-window-frame-focus))))))))

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Dmitry Yaitskov]

Thanks, your patch fixes the problem!

Shenghuo ZHU <zsh@cs.rochester.edu> wrote:

> >>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> 
> LMI> Colin Rafferty <colin@xemacs.org> writes:
> >> There were some definite changes in redisplay.  I think that it
> >> doesn't redisplay until later than it used to.  We also do pixel-based
> >> scrolling, which screwed up shrink-window-if-larger-than-buffer' (a
> >> fix to which I am testing).
> >> 
> >> Do you do anything with window-pixel-edges?  I had to remove that.
> 
> LMI> I don't use that -- just `window-height', I think.  Perhaps that is
> LMI> being computed later than it used to be?
> 
> As I remember, just before 5.8.3 was released, it was OK.  The
> difference is that switch-to-buffer was changed to set-buffer in
> gnus-configure-windows. I don't know the reason, but the attached
> patch works for me (in XEmacs 21.2.32). This patch has not been
> applied to the CVS repository. If it really works, could anyone tell
> me the reason?

-- 
Cheers,
-Dima.

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[Lars Magne Ingebrigtsen]

Shenghuo ZHU <zsh@cs.rochester.edu> writes:

> As I remember, just before 5.8.3 was released, it was OK.  The
> difference is that switch-to-buffer was changed to set-buffer in
> gnus-configure-windows. I don't know the reason, but the attached
> patch works for me (in XEmacs 21.2.32). This patch has not been
> applied to the CVS repository. If it really works, could anyone tell
> me the reason?

I've applied the patch.

The reason I made the change is that the `switch-to-buffer'-ness of
the function made things change in the buffer ordering so that `C-x b' 
(in some cases) worked in a quirky fashion after Gnus arranged the
windows.  

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: gnus 5.8.4 - display bug (same as in 5.8.3)

[David Hedbor]

Dmitry Yaitskov <dimas@home.com> writes:

> Hi,
> 
> This is IIRC the same (or very similar) bug which prevented me from
> ever using 5.8.3: Enter a newsgroup (the *Summary* buffer), w/out

[bug desc snipped]

That is exactly the bug, yes.

-- 
[ Below is a random fortune, which is unrelated to the above message. ]
The only "intuitive" interface is the nipple. After that, it's all learned.
(Bruce Ediger, bediger@teal.csn.org, in comp.os.linux.misc, on X interfaces.)