bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[Michael Welsh Duggan]

[-- Attachment #1: Type: text/plain, Size: 276 bytes --]


echo "(setq default-frame-alist '((fullscreen . fullheight)))" > /tmp/foo.el
emacs -Q --daemon -l /tmp/foo.el
emacsclient -c -n

Using the above recipe, I get a frame width of 78 characters instead of
80 characters.  The information from (window--frame-dump) is as follows:


[-- Attachment #2: frame.txt --]
[-- Type: text/plain, Size: 892 bytes --]

frame pixel: 658 x 971   cols/lines: 82 x 54   units: 8 x 17
frame text pixel: 624 x 933   cols/lines: 78 x 54
tool: 36  scroll: 16/0  fringe: 16  border: 1  right: 0  bottom: 0

#<window 3 on *scratch*>   parent: nil
pixel left: 0   top: 36   size: 656 x 916   new: 0
char left: 0   top: 3   size: 82 x 53   new: 0
normal: 1.0 x 1.0   new: 0
body pixel: 624 x 899   char: 78 x 52
width left fringe: 8  left margin: 0  right margin: 0
width right fringe: 8  scroll-bar: 16  divider: 0
height header-line: 0  mode-line: 17  divider: 0

#<window 4 on  *Minibuf-0*>   parent: nil
pixel left: 0   top: 952   size: 656 x 17   new: 0
char left: 0   top: 56   size: 656 x 1   new: 0
normal: 1.0 x 1.0   new: 0
body pixel: 624 x 17   char: 78 x 1
width left fringe: 8  left margin: 0  right margin: 0
width right fringe: 8  scroll-bar: 16  divider: 0
height header-line: 0  mode-line: 0  divider: 0


[-- Attachment #3: Type: text/plain, Size: 781 bytes --]


In GNU Emacs 25.0.50.1 (x86_64-unknown-linux-gnu, X toolkit)
 of 2015-01-15 on watermonitor.yellow.cert.org
Repository revision: c263a40a9ac76a87b919ca8d425736797c3c399c
Windowing system distributor `Red Hat, Inc.', version 11.0.11500000
System Description:	Red Hat Enterprise Linux Server release 6.6 (Santiago)

Configured using:
 `configure --without-toolkit-scroll-bars --with-x-toolkit=lucid
 --with-wide-int --prefix=/home/mwd
 PKG_CONFIG_PATH=/home/mwd/lib/pkgconfig:/usr/lib/pkgconfig:/usr/share/pkgconfig'

Configured features:
XPM JPEG TIFF GIF PNG IMAGEMAGICK SOUND GPM DBUS NOTIFY LIBSELINUX
GNUTLS LIBXML2 FREETYPE M17N_FLT LIBOTF XFT ZLIB

Important settings:
  value of $LANG: en_US.UTF-8
  locale-coding-system: utf-8-unix

-- 
Michael Welsh Duggan
(mwd@cert.org)

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[martin rudalics]

[-- Attachment #1: Type: text/plain, Size: 548 bytes --]

 > echo "(setq default-frame-alist '((fullscreen . fullheight)))" > /tmp/foo.el
 > emacs -Q --daemon -l /tmp/foo.el
 > emacsclient -c -n
 >
 > Using the above recipe, I get a frame width of 78 characters instead of
 > 80 characters.  The information from (window--frame-dump) is as follows:

Thanks.  I haven't looked into your recipe yet.  Does the attached patch
improve anything?  If it doesn't apply right away just tell me and don't
bother any further.  I'm working on this problem currently and have
found a couple of related issues.

martin

[-- Attachment #2: fullscreen.diff --]
[-- Type: text/plain, Size: 4985 bytes --]

diff --git a/lisp/frame.el b/lisp/frame.el
index 1d5bbf2..c18e4be 100644
--- a/src/frame.c
+++ b/src/frame.c
@@ -152,10 +152,13 @@ get_frame_param (register struct frame *frame, Lisp_Object prop)
 bool
 frame_inhibit_resize (struct frame *f, bool horizontal, Lisp_Object parameter)
 {
+  Lisp_Object fullscreen = get_frame_param (f, Qfullscreen);
+
   return (EQ (frame_inhibit_implied_resize, Qt)
 	  || (CONSP (frame_inhibit_implied_resize)
 	      && !NILP (Fmemq (parameter, frame_inhibit_implied_resize)))
-	  || !NILP (get_frame_param (f, Qfullscreen))
+	  || (horizontal && !EQ (fullscreen, Qnil) && !EQ (fullscreen, Qfullheight))
+	  || (!horizontal && !EQ (fullscreen, Qnil) && !EQ (fullscreen, Qfullwidth))
 	  || FRAME_TERMCAP_P (f) || FRAME_MSDOS_P (f));
 }
 
@@ -1009,7 +1012,8 @@ affects all frames on the same terminal device.  */)
   {
     int width, height;
     get_tty_size (fileno (FRAME_TTY (f)->input), &width, &height);
-    adjust_frame_size (f, width, height - FRAME_MENU_BAR_LINES (f), 5, 0, Qnil);
+    adjust_frame_size (f, width, height - FRAME_MENU_BAR_LINES (f),
+		       5, 0, Qterminal_frame);
   }
 
   adjust_frame_glyphs (f);
@@ -3207,14 +3211,14 @@ x_set_frame_parameters (struct frame *f, Lisp_Object alist)
 	   that here since otherwise a size change implied by an
 	   intermittent font change may get lost as in Bug#17142.  */
 	if (!width_change)
-	  width = (f->new_width
+	  width = ((f->can_x_set_window_size && f->new_width)
 		   ? (f->new_pixelwise
 		      ? f->new_width
 		      : (f->new_width * FRAME_COLUMN_WIDTH (f)))
 		   : FRAME_TEXT_WIDTH (f));
 
 	if (!height_change)
-	  height = (f->new_height
+	  height = ((f->can_x_set_window_size && f->new_height)
 		    ? (f->new_pixelwise
 		       ? f->new_height
 		       : (f->new_height * FRAME_LINE_HEIGHT (f)))
@@ -4828,6 +4832,8 @@ syms_of_frame (void)
   DEFSYM (Qset_window_configuration, "set-window-configuration");
   DEFSYM (Qx_create_frame_1, "x-create-frame-1");
   DEFSYM (Qx_create_frame_2, "x-create-frame-2");
+  DEFSYM (Qtip_frame, "tip-frame");
+  DEFSYM (Qterminal_frame, "terminal-frame");
 
 #ifdef HAVE_NS
   DEFSYM (Qns_parse_geometry, "ns-parse-geometry");
diff --git a/src/w32fns.c b/src/w32fns.c
index 2dd92ff..1db47f2 100644
--- a/src/w32fns.c
+++ b/src/w32fns.c
@@ -1745,7 +1745,8 @@ x_change_tool_bar_height (struct frame *f, int height)
   f->n_tool_bar_rows = 0;
 
   adjust_frame_size (f, -1, -1,
-		     (!f->tool_bar_redisplayed_once ? 1
+		     ((!f->tool_bar_redisplayed_once
+		       && NILP (get_frame_param (f, Qfullscreen))) ? 1
 		      : (old_height == 0 || height == 0) ? 2
 		      : 4),
 		     0, Qtool_bar_lines);
@@ -5786,7 +5787,7 @@ x_create_tip_frame (struct w32_display_info *dpyinfo,
   SET_FRAME_COLS (f, 0);
   SET_FRAME_LINES (f, 0);
   adjust_frame_size (f, width * FRAME_COLUMN_WIDTH (f),
-		     height * FRAME_LINE_HEIGHT (f), 0, 1, Qnil);
+		     height * FRAME_LINE_HEIGHT (f), 0, 1, Qtip_frame);
 
   /* Add `tooltip' frame parameter's default value. */
   if (NILP (Fframe_parameter (frame, Qtooltip)))
diff --git a/src/xfns.c b/src/xfns.c
index 936c769..c1a9e6c 100644
--- a/src/xfns.c
+++ b/src/xfns.c
@@ -1079,7 +1079,7 @@ x_set_tool_bar_lines (struct frame *f, Lisp_Object value, Lisp_Object oldval)
 void
 x_change_tool_bar_height (struct frame *f, int height)
 {
-#ifdef USE_GTK
+#if defined (USE_GTK) || defined (HAVE_NS)
   FRAME_TOOL_BAR_LINES (f) = 0;
   FRAME_TOOL_BAR_HEIGHT (f) = 0;
   if (height)
@@ -1100,6 +1100,7 @@ x_change_tool_bar_height (struct frame *f, int height)
   int unit = FRAME_LINE_HEIGHT (f);
   int old_height = FRAME_TOOL_BAR_HEIGHT (f);
   int lines = (height + unit - 1) / unit;
+  Lisp_Object fullscreen;
 
   /* Make sure we redisplay all windows in this frame.  */
   windows_or_buffers_changed = 60;
@@ -1131,7 +1132,10 @@ x_change_tool_bar_height (struct frame *f, int height)
   f->n_tool_bar_rows = 0;
 
   adjust_frame_size (f, -1, -1,
-		     (!f->tool_bar_redisplayed_once ? 1
+		     ((!f->tool_bar_redisplayed_once
+		       && (NILP (fullscreen =
+				 get_frame_param (f, Qfullscreen))
+			   || EQ (fullscreen, Qfullwidth))) ? 1
 		      : (old_height == 0 || height == 0) ? 2
 		      : 4),
 		     0, Qtool_bar_lines);
diff --git a/src/xterm.c b/src/xterm.c
index 3955d02..971f342 100644
--- a/src/xterm.c
+++ b/src/xterm.c
@@ -9343,13 +9343,22 @@ x_check_fullscreen (struct frame *f)
           break;
         case FULLSCREEN_WIDTH:
           width = x_display_pixel_width (dpyinfo);
-          break;
+	  height = height + FRAME_MENUBAR_HEIGHT (f);
+	break;
         case FULLSCREEN_HEIGHT:
           height = x_display_pixel_height (dpyinfo);
         }
 
       XResizeWindow (FRAME_X_DISPLAY (f), FRAME_OUTER_WINDOW (f),
                      width, height);
+
+      if (FRAME_VISIBLE_P (f))
+	x_wait_for_event (f, ConfigureNotify);
+      else
+	{
+	  change_frame_size (f, width, height, false, true, false, true);
+	  x_sync (f);
+	}
     }
 }
 

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[Michael Welsh Duggan]

martin rudalics <rudalics@gmx.at> writes:

>> echo "(setq default-frame-alist '((fullscreen . fullheight)))" > /tmp/foo.el
>> emacs -Q --daemon -l /tmp/foo.el
>> emacsclient -c -n
>>
>> Using the above recipe, I get a frame width of 78 characters instead of
>> 80 characters.  The information from (window--frame-dump) is as follows:
>
> Thanks.  I haven't looked into your recipe yet.  Does the attached patch
> improve anything?  If it doesn't apply right away just tell me and don't
> bother any further.  I'm working on this problem currently and have
> found a couple of related issues.

That patch does seem to solve my problem.

-- 
Michael Welsh Duggan
(mwd@cert.org)

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[martin rudalics]

> That patch does seem to solve my problem.

OK.  Use it and complain immediately when you find more problems.
I plan to commit a modified version of it in a couple of days.

Thanks, martin

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[martin rudalics]

 > That patch does seem to solve my problem.

Can you please try how current trunk/master behaves in this regard?

Thanks, martin

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[Michael Welsh Duggan]

martin rudalics <rudalics@gmx.at> writes:

>> That patch does seem to solve my problem.
>
> Can you please try how current trunk/master behaves in this regard?

Things seem to work correctly in a recently pulled master.  Thank you!

-- 
Michael Welsh Duggan
(md5i@md5i.com)

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[Stephen Berman]

On Sun, 08 Feb 2015 00:54:37 -0500 Michael Welsh Duggan <mwd@md5i.com> wrote:

> martin rudalics <rudalics@gmx.at> writes:
>
>>> That patch does seem to solve my problem.
>>
>> Can you please try how current trunk/master behaves in this regard?
>
> Things seem to work correctly in a recently pulled master.  Thank you!

FTR, this patch has evidently also fixed a curious bug I was about to
report, until I pulled the latest master, rebuilt and failed to
reproduce it.  The bug was that the combination of (i) adding the frame
parameter (fullscreen . fullheight), (ii) customizing tool-bar-style to
'left and (iii) calling tool-bar-add-item or tool-bar-add-item-from-menu
resulted in lines longer than window-width wrapping one column too
soon.  Again, with current master, this bug is no longer reproducible.
So thanks for the prescience (even if unintended)!

Steve Berman

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[martin rudalics]

 > The bug was that the combination of (i) adding the frame
 > parameter (fullscreen . fullheight),

... adding to `default-frame-alist', `initial-frame-alist'?

 > (ii) customizing tool-bar-style to
 > 'left

`tool-bar-style' or the `tool-bar-position' parameter?

 > and (iii) calling tool-bar-add-item or tool-bar-add-item-from-menu
 > resulted in lines longer than window-width wrapping one column too
 > soon.  Again, with current master, this bug is no longer reproducible.
 > So thanks for the prescience (even if unintended)!

Sheer luck, probably.  Does the problem already happen with Emacs 24.5?
And what happens with a fullwidth frame and the toolbar orthogonally to
where it is now?

martin

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[Stephen Berman]

On Mon, 09 Feb 2015 10:31:19 +0100 martin rudalics <rudalics@gmx.at> wrote:

>> The bug was that the combination of (i) adding the frame
>> parameter (fullscreen . fullheight),
>
> ... adding to `default-frame-alist', `initial-frame-alist'?

I used modify-frame-parameters.

>> (ii) customizing tool-bar-style to
>> 'left
>
> `tool-bar-style' or the `tool-bar-position' parameter?

The latter, sorry for the confusion.

>> and (iii) calling tool-bar-add-item or tool-bar-add-item-from-menu
>> resulted in lines longer than window-width wrapping one column too
>> soon.  Again, with current master, this bug is no longer reproducible.
>> So thanks for the prescience (even if unintended)!
>
> Sheer luck, probably.  Does the problem already happen with Emacs 24.5?

It does not happen with my latest build from the emacs-24 branch.  (That
does not use a toolkit scroll bar, unlike my builds from master, though I
guess that isn't relevant here.)

> And what happens with a fullwidth frame and the toolbar orthogonally to
> where it is now?

Do you mean using the parameter (fullscreen . fullboth) and tool bar on
the left?  If so, the result is the same in master with and without you
recent patch: there's a bit of space between the last character and the
right fringe, but less than one column's worth.

Although, with your patch, I no longer observe the problem, in case you
want to do some ex post facto forensics, below is the minimal recipe I
came up with to show the problem.  Do emacs -Q in a build from master
without your patch, evaluate the code and then type `M-x srb-test'.

Steve Berman

(defun srb-test ()
  "Test display of rightmost column."
  (interactive)
  (let ((buf (generate-new-buffer "*srb test*")))
    (with-current-buffer buf
      (insert "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.")
      (srb-test-mode))
    (switch-to-buffer buf)))

(defvar srb-test-toolbar
  (let ((tool-bar-map (make-sparse-keymap)))
    (tool-bar-add-item nil nil nil)
    tool-bar-map))

(defvar srb-test-mode-map
  (let ((map (make-sparse-keymap)))
    (define-key map [menu-bar] (make-sparse-keymap))
    map)
  "Keymap for `srb-test-mode'.")

(define-derived-mode srb-test-mode nil "srb test"
  "Mode to test display of rightmost column."
  (modify-frame-parameters nil '((fullscreen . fullheight)))
  (custom-set-variables '(tool-bar-position (quote left)))
  (setq-local tool-bar-map srb-test-toolbar))

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[martin rudalics]

 > It does not happen with my latest build from the emacs-24 branch.  (That
 > does not use a toolkit scroll bar, unlike my builds from master, though I
 > guess that isn't relevant here.)

I think so too.  The problem is the toolbar which gets resized _after_
the frame was made fullheight and before my last patch this (i.e., being
fullheight) inhibited changing the frame's width as well.  For example,
if you had set the text width to 80 columns and enlarged the tool bar by
say 4 columns that would have consumed 4 columns from the text and not
enlarged the combined width of frame plus toolbar by 4 columns.  That's
what Michael detected and what I tried to fix lately.

In fact, the problem was introduced by an earlier fix for the orthogonal
problem where I tried to keep a frame fullheight/-width when changing
the size of the toolbar on the side of the largest extension.  So you
should not see it on Emacs 24.4/5.

 >> And what happens with a fullwidth frame and the toolbar orthogonally to
 >> where it is now?
 >
 > Do you mean using the parameter (fullscreen . fullboth) and tool bar on
 > the left?

I meant (fullscreen . fullwidth) and the tool bar on top.  But that's
irrelevant wrt to how many characters fit on a line, so drop it.

 > If so, the result is the same in master with and without you
 > recent patch: there's a bit of space between the last character and the
 > right fringe, but less than one column's worth.

Probably because a fullboth frame's text width is not a multiple of the
character width.  What does evaluating (window--dump-frame) give on that
frame?  The result is in a buffer called *window-frame-dump*.  If the
second value I marked below is not the product of the first and third
value I marked you should see that.

frame pixel: 1680 x 984   cols/lines: 210 x 60   units: 8 x 16
                                                         ^
frame text pixel: 1648 x 968   cols/lines: 206 x 60
                   ^^^^                     ^^^
tool: 0  scroll: 16/16  fringe: 16  border: 0  right: 6  bottom: 6

 > Although, with your patch, I no longer observe the problem, in case you
 > want to do some ex post facto forensics, below is the minimal recipe I
 > came up with to show the problem.  Do emacs -Q in a build from master
 > without your patch, evaluate the code and then type `M-x srb-test'.

I can see it with that recipe.

Thanks, martin

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[Stephen Berman]

On Mon, 09 Feb 2015 19:41:54 +0100 martin rudalics <rudalics@gmx.at> wrote:

>> Do you mean using the parameter (fullscreen . fullboth) and tool bar on
>> the left?
>
> I meant (fullscreen . fullwidth) and the tool bar on top.  But that's
> irrelevant wrt to how many characters fit on a line, so drop it.
>
>> If so, the result is the same in master with and without you
>> recent patch: there's a bit of space between the last character and the
>> right fringe, but less than one column's worth.
>
> Probably because a fullboth frame's text width is not a multiple of the
> character width.  What does evaluating (window--dump-frame) give on that
> frame?  The result is in a buffer called *window-frame-dump*.  If the
> second value I marked below is not the product of the first and third
> value I marked you should see that.
>
> frame pixel: 1680 x 984   cols/lines: 210 x 60   units: 8 x 16
>                                                         ^
> frame text pixel: 1648 x 968   cols/lines: 206 x 60
>                   ^^^^                     ^^^
> tool: 0  scroll: 16/16  fringe: 16  border: 0  right: 6  bottom: 6

Indeed it's not (i.e. 9x129=1161):

frame pixel: 1199 x 1004   cols/lines: 133 x 52   units: 9 x 19
frame text pixel: 1167 x 1004   cols/lines: 129 x 52
tool: 0  scroll: 16/0  fringe: 16  border: 0  right: 0  bottom: 0

Steve Berman

bug#19619: 25.0.50; Fullheight frames in daemonized emacsen are wrong width

[martin rudalics]

> Things seem to work correctly in a recently pulled master.  Thank you!

Closing.

Thank you for the report, martin