Wrong frame with recent trunk [MSYS2-MinGW64]

Wrong frame with recent trunk [MSYS2-MinGW64]

[Angelo Graziosi]

With recent trunk (say rev. >= r117564), my MSYS2-mingw64 (Win7 64) 
Emacs builds show a strange behaviour : each time I start Emacs, the 
frame is 2 line shorter in height. For example if I start with a frame 
of 36 line, the next start it is 34, then 32, etc.

I can reproduce this with this minimal init.el [*]

$ cat /c/Users/pippo/AppData/Roaming/.emacs.d/init.el
;; Instead to save in ~/.emacs.desktop
(setq desktop-base-file-name "~/.emacs.d/desktop")

;; Instead to save in ~/.emacs.desktop.lock
(setq desktop-base-lock-name "~/.emacs.d/desktop.lock")

;; Save desktop
(desktop-save-mode t)


---
[*] From Emacs-Windows point of view, "~" is 
"/c/Users/pippo/AppData/Roaming" (on Win7)

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Dmitry Antipov]

On 07/24/2014 01:35 AM, Angelo Graziosi wrote:

> With recent trunk (say rev. >= r117564), my MSYS2-mingw64 (Win7 64)
> Emacs builds show a strange behavior : each time I start Emacs, the
> frame is 2 line shorter in height. For example if I start with
> a frame of 36 line, the next start it is 34, then 32, etc.

Hopefully fixed, please try revision >= 117567.

Can you also try that the following:

(set-frame-height nil (1+ (frame-height)))
(set-frame-height nil (1- (frame-height)))
(set-frame-size nil (1+ (frame-width)) (1+ (frame-height)))
(set-frame-size nil (1- (frame-width)) (1- (frame-height)))

works as expected on MS-Windows?

Dmitry

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Angelo Graziosi]

Il 24/07/2014 07:52, Dmitry Antipov ha scritto:
>
> Hopefully fixed, please try revision >= 117567.

Yes, it seems so..

>
> Can you also try that the following:
>
> (set-frame-height nil (1+ (frame-height)))
> (set-frame-height nil (1- (frame-height)))
> (set-frame-size nil (1+ (frame-width)) (1+ (frame-height)))
> (set-frame-size nil (1- (frame-width)) (1- (frame-height)))

Sorry, may you clarify?

Should I put that code in an init.el (.emacs etc.) file?

> works as expected on MS-Windows?

What should I expect?

Ciao,
  Angelo.

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Dmitry Antipov]

On 07/24/2014 04:40 PM, Angelo Graziosi wrote:

>> Can you also try that the following:
>>
>> (set-frame-height nil (1+ (frame-height)))
>> (set-frame-height nil (1- (frame-height)))
>> (set-frame-size nil (1+ (frame-width)) (1+ (frame-height)))
>> (set-frame-size nil (1- (frame-width)) (1- (frame-height)))
>
> Sorry, may you clarify?
>
> Should I put that code in an init.el (.emacs etc.) file?

No, just copy & paste them into *scratch*, then go to end
of each line and type C-x C-e.

>> works as expected on MS-Windows?
>
> What should I expect?

(set-frame-height nil (1+ (frame-height)))
==> frame height should be increased by one character row

(set-frame-height nil (1- (frame-height)))
==> likewise but decreased

(set-frame-size nil (1+ (frame-width)) (1+ (frame-height)))
==> width should be increased by one character column and
     height should be increased by one character row

 >> (set-frame-size nil (1- (frame-width)) (1- (frame-height)))
==> likewise but decreased

Dmitry

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Angelo Graziosi]

Il 24/07/2014 18:22, Dmitry Antipov ha scritto:

>
> No, just copy & paste them into *scratch*, then go to end
> of each line and type C-x C-e.

Oh, I did this but didn't notice the effect without something to 
compare.. Now I resize the window of a running application to the same 
size of Emacs frame and noticed the result...

>
>>> works as expected on MS-Windows?
>>
>> What should I expect?
>
> (set-frame-height nil (1+ (frame-height)))
> ==> frame height should be increased by one character row
>
> (set-frame-height nil (1- (frame-height)))
> ==> likewise but decreased
>
> (set-frame-size nil (1+ (frame-width)) (1+ (frame-height)))
> ==> width should be increased by one character column and
>      height should be increased by one character row
>
>  >> (set-frame-size nil (1- (frame-width)) (1- (frame-height)))
> ==> likewise but decreased


Yes, unless I have been drinking a lot of wine.. sob..., all that works 
as expected also on Windows... ;-)


Ciao,
Angelo.

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Angelo Graziosi]

It seems the the issue has been reintroduced in some manner..

Revision 117568 looks ok: when I restart Emacs the frame has the same 
position, width and height of previous session.

With the rev. 117589, restarting Emacs produces a frame with the same 
position and width but the height increases od about 2 line, so after 
one restarts 2, 3... times, Emacs frame is too long, out the screen..

Maybe recent Martin's changes about horizontal scroll bar causes this?

Thanks,
  angelo.

Il 24/07/2014 18:44, Angelo Graziosi ha scritto:
>
>
> Il 24/07/2014 18:22, Dmitry Antipov ha scritto:
>
>>
>> No, just copy & paste them into *scratch*, then go to end
>> of each line and type C-x C-e.
>
> Oh, I did this but didn't notice the effect without something to
> compare.. Now I resize the window of a running application to the same
> size of Emacs frame and noticed the result...
>
>>
>>>> works as expected on MS-Windows?
>>>
>>> What should I expect?
>>
>> (set-frame-height nil (1+ (frame-height)))
>> ==> frame height should be increased by one character row
>>
>> (set-frame-height nil (1- (frame-height)))
>> ==> likewise but decreased
>>
>> (set-frame-size nil (1+ (frame-width)) (1+ (frame-height)))
>> ==> width should be increased by one character column and
>>      height should be increased by one character row
>>
>>  >> (set-frame-size nil (1- (frame-width)) (1- (frame-height)))
>> ==> likewise but decreased
>
>
> Yes, unless I have been drinking a lot of wine.. sob..., all that works
> as expected also on Windows... ;-)
>
>
> Ciao,
> Angelo.
>

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Eli Zaretskii]

> Date: Mon, 28 Jul 2014 00:16:55 +0200
> From: Angelo Graziosi <angelo.graziosi@alice.it>
> Cc: rudalics@gmx.at, Emacs developers <emacs-devel@gnu.org>
> 
> With the rev. 117589, restarting Emacs produces a frame with the same 
> position and width but the height increases od about 2 line, so after 
> one restarts 2, 3... times, Emacs frame is too long, out the screen..

Not reproducible here with a native w32 build of r117589.

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Eli Zaretskii]

> Date: Mon, 28 Jul 2014 09:00:46 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: rudalics@gmx.at, dmantipov@yandex.ru, emacs-devel@gnu.org
> 
> > Date: Mon, 28 Jul 2014 00:16:55 +0200
> > From: Angelo Graziosi <angelo.graziosi@alice.it>
> > Cc: rudalics@gmx.at, Emacs developers <emacs-devel@gnu.org>
> > 
> > With the rev. 117589, restarting Emacs produces a frame with the same 
> > position and width but the height increases od about 2 line, so after 
> > one restarts 2, 3... times, Emacs frame is too long, out the screen..
> 
> Not reproducible here with a native w32 build of r117589.

Sorry, that was inaccurate.

First, you didn't say what you mean by "restarting Emacs".  I
originally thought just "emacs -Q" will do, but that comes up with the
same size every time.  My current guess is that you have
desktop-save-mode enabled, and you invoke Emacs without -Q, is that
correct?

Second, even with desktop-save-mode, I get a 2-line increment only one
time, the first time I start Emacs after saving the desktop.  All the
subsequent invocations keep that two-line-larger size.

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Angelo Graziosi]

Il 28/07/2014 08:15, Eli Zaretskii ha scritto:

>>
>>>
>>> With the rev. 117589, restarting Emacs produces a frame with the same
>>> position and width but the height increases od about 2 line, so after
>>> one restarts 2, 3... times, Emacs frame is too long, out the screen..
>>
>> Not reproducible here with a native w32 build of r117589.
>
> Sorry, that was inaccurate.
> First, you didn't say what you mean by "restarting Emacs".  I
> originally thought just "emacs -Q" will do, but that comes up with the
> same size every time.  My current guess is that you have
> desktop-save-mode enabled, and you invoke Emacs without -Q, is that
> correct?
>
> Second, even with desktop-save-mode, I get a 2-line increment only one
> time, the first time I start Emacs after saving the desktop.  All the
> subsequent invocations keep that two-line-larger size.
>

There are still issues with trunk rev. 117599.

Now restarting Emacs (with a link pointing to runemacs), the frame 
conserve the position of previous session BUT NOT the WIDTH nor the HEIGHT.

Suppose to have

$ cat /c/Users/pippo/AppData/Roaming/.emacs.d/init.el
;; Instead to save in ~/.emacs.desktop
(setq desktop-base-file-name "~/.emacs.d/desktop")

;; Instead to save in ~/.emacs.desktop.lock
(setq desktop-base-lock-name "~/.emacs.d/desktop.lock")

;; Save desktop
(desktop-save-mode t)


and


$ ls -lrta /c/Users/pippo/AppData/Roaming/.emacs.d/
totale 9
-rw-r--r-- 1 pippo None 230 28 lug 09.28 init.el
drwxr-xr-x 1 pippo None   0 28 lug 11.54 ..
drwxr-xr-x 1 pippo None   0 28 lug 12.02 .


Now start Emacs with a link pointing to runemacs (the link created by addpm)

Emacs opens with almost a square frame. Change its position (for example 
centered on Windows desktop) and resize it (for example, set it as a 
rectangular frame).

Now close it (click on the 'x' in the upper right corner). It asks if 
you want to save the desktop. 'Yes'

Now restart Emacs with the same link. The position is the same of 
previous session (centered on Windows desktop) but the size is not 
rectangular as you have left it but the same original square size.

This occurs in rev. 117599.

Just out of curiosity... How you build Emacs? I build always in a 
clean[*] tree with:

configure --prefix=/Emacs.app --with-wide-int
  --build=x86_64-w64-mingw32 --without-imagemagick
  'CFLAGS=-I/mingw64/include/noX -Ofast -g0 -pipe' LDFLAGS=-pipe

make -j3
...


Ciao,
  Angelo.

---
[*] I have a local repository created with:

   $ bzr checkout --lightweight 
http://bzr.savannah.gnu.org/r/emacs/trunk emacs-trunk

I update it with 'bzr up'. I DO NOT use it to build but I use a copy:

   rsync -av --exclude=.bzr* --exclude=.git* --exclude=.gdb* --delete 
/c/cygwin64/home/pippo/work/emacs-trunk/ /tmp/emacs/

then in a mingw64_shell.bat shell:

   $ cd /tmp
   $ .... build-emacs.sh

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Eli Zaretskii]

> Date: Mon, 28 Jul 2014 12:26:47 +0200
> From: Angelo Graziosi <angelo.graziosi@alice.it>
> CC: rudalics@gmx.at, dmantipov@yandex.ru, emacs-devel@gnu.org
> 
> There are still issues with trunk rev. 117599.

Please submit a proper bug report to the bug tracker.  I think we've
discussed this issue more than enough here, and if it's still a
problem, we need to treat it as such.

> Just out of curiosity... How you build Emacs?

Just "make -j6" (which will run autoconf if needed).

> I build always in a clean[*] tree with:
> 
> configure --prefix=/Emacs.app --with-wide-int
>   --build=x86_64-w64-mingw32 --without-imagemagick
>   'CFLAGS=-I/mingw64/include/noX -Ofast -g0 -pipe' LDFLAGS=-pipe
> 
> make -j3
> ...
> 
> 
> Ciao,
>   Angelo.
> 
> ---
> [*] I have a local repository created with:
> 
>    $ bzr checkout --lightweight 
> http://bzr.savannah.gnu.org/r/emacs/trunk emacs-trunk
> 
> I update it with 'bzr up'. I DO NOT use it to build but I use a copy:
> 
>    rsync -av --exclude=.bzr* --exclude=.git* --exclude=.gdb* --delete 
> /c/cygwin64/home/pippo/work/emacs-trunk/ /tmp/emacs/
> 
> then in a mingw64_shell.bat shell:
> 
>    $ cd /tmp
>    $ .... build-emacs.sh

I don't have time for all this complexity, and cannot use lightweight
checkouts to begin with.  Bootstrapping Emacs every time just wastes
time and as a nasty side effect removes all the previous *.exe files
that I could otherwise use to kind-of bisect problems.  I never
understood people who always bootstrap.

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[martin rudalics]

 > With the rev. 117589, restarting Emacs produces a frame with the same position and width but the height increases od about 2 line, so after one restarts 2, 3... times, Emacs frame is too long, out the screen..
 >
 > Maybe recent Martin's changes about horizontal scroll bar causes this?

Actually it was a leftover from Dmitry's previous change.  Since the
text height of a frame no more includes the top margin, adding its size
to it caused the frame height to increase.  Should be fixed now.

Thanks for the prompt reaction, martin

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[martin rudalics]

 > Now restarting Emacs (with a link pointing to runemacs), the frame conserve the position of previous session BUT NOT the WIDTH nor the HEIGHT.
[...]
 > Now restart Emacs with the same link. The position is the same of previous session (centered on Windows desktop) but the size is not rectangular as you have left it but the same original square size.

This was due to an undocumented change in revision 117591.  I don't know
what Paul wanted to do there so I simply reverted it.  Paul?

Thanks, martin

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Angelo Graziosi]

It seems the issue ha been reintroduced, at least in rev. 117616: the 
frame does not restore the size it had in the last session.

I wonder it this has to do with

2014-07-31  Paul Eggert  <eggert@cs.ucla.edu>

         [...]

         * frame.c (x_set_frame_parameters): Don't use uninitialized locals.
         Without this change, the code can access the local variable 'width'
         even when it has not been initialized, and likewise for 'height';
         in either case this leads to undefined behavior.


Paul, any comment?

Angelo

Il 28/07/2014 19:26, martin rudalics ha scritto:
>  > Now restarting Emacs (with a link pointing to runemacs), the frame
> conserve the position of previous session BUT NOT the WIDTH nor the HEIGHT.
> [...]
>  > Now restart Emacs with the same link. The position is the same of
> previous session (centered on Windows desktop) but the size is not
> rectangular as you have left it but the same original square size.
>
> This was due to an undocumented change in revision 117591.  I don't know
> what Paul wanted to do there so I simply reverted it.  Paul?
>
> Thanks, martin

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[chad]

On 31 Jul 2014, at 15:53, Angelo Graziosi <angelo.graziosi@alice.it> wrote:

> It seems the issue ha been reintroduced, at least in rev. 117616: the frame does not restore the size it had in the last session.

FWIW, I also saw this when I tried HEAD for a bit, but don't see
it in the 24.4 rc branch, which is my normal emacs right now.

~Chad

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Angelo Graziosi]

Il 01/08/2014 01:05, chad ha scritto:
>
> On 31 Jul 2014, at 15:53, Angelo Graziosi <angelo.graziosi@alice.it> wrote:
>
>> It seems the issue ha been reintroduced, at least in rev. 117616: the frame does not restore the size it had in the last session.
>
> FWIW, I also saw this when I tried HEAD for a bit, but don't see
> it in the 24.4 rc branch, which is my normal emacs right now.

Because Paul's change regards trunk (i.e. Head) and not 24.4 branch, i 
think...

Angelo

Re: Wrong frame with recent trunk [MSYS2-MinGW64]

[Paul Eggert]

[-- Attachment #1: Type: text/plain, Size: 128 bytes --]

Angelo Graziosi wrote:
> Paul, any comment?

Sorry, I had a brain freeze in that patch.  Please try trunk bzr 117617 
or later.

[-- Attachment #2: width.patch --]
[-- Type: text/plain, Size: 540 bytes --]

=== modified file 'src/frame.c'
--- src/frame.c	2014-07-31 13:55:12 +0000
+++ src/frame.c	2014-07-31 23:53:14 +0000
@@ -3198,9 +3198,7 @@ x_set_frame_parameters (struct frame *f,
 
     XSETFRAME (frame, f);
 
-    if (((width_change && width != FRAME_TEXT_WIDTH (f))
-	 || (height_change && height != FRAME_TEXT_HEIGHT (f)))
-	&& (f->new_height || f->new_width))
+    if (width_change || height_change)
       {
 	/* If necessary provide default values for HEIGHT and WIDTH.  Do
 	   that here since otherwise a size change implied by an