bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[Drew Adams]

emacs -Q

M-: (set-face-attribute 'default nil :font "-outline-Lucida Console-normal-normal-normal-mono-14-*-*-*-c-*-iso8859-1")

That correctly sets the font for the current frame.  However, it does
NOT set the font for future frames, as the doc says it should and as it
does in previous Emacs versions (e.g. 24.3).

Similarly, this does not work anymore:

M-: (set-face-attribute 'default t :font "-outline-Lucida Console-normal-normal-normal-mono-14-*-*-*-c-*-iso8859-1")

In GNU Emacs 24.3.50.1 (i686-pc-mingw32)
 of 2014-01-17 on ODIEONE
Bzr revision: 116059 michael.albinus@gmx.de-20140117182409-gz695p2c951idlee
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --prefix=/c/Devel/emacs/binary --enable-checking=yes,glyphs
 'CFLAGS=-O0 -g3' LDFLAGS=-Lc:/Devel/emacs/lib
 CPPFLAGS=-Ic:/Devel/emacs/include'

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[Glenn Morris]

Works for me.

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[Juanma Barranquero]

> Works for me.

Doesn't work for me. Perhaps is Windows-specific.

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Fri, 24 Jan 2014 18:43:37 +0100
> Cc: 16529@debbugs.gnu.org
> 
> > Works for me.
> 
> Doesn't work for me. Perhaps is Windows-specific.

Could you please bisect this?  I have no idea how does the font
parameter get propagated to other frames (it's not via
default-frame-alist), so I don't know where to look for the cause of
this problem.

Thanks.

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[Eli Zaretskii]

> Date: Sat, 25 Jan 2014 14:01:55 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: 16529@debbugs.gnu.org
> 
> > From: Juanma Barranquero <lekktu@gmail.com>
> > Date: Fri, 24 Jan 2014 18:43:37 +0100
> > Cc: 16529@debbugs.gnu.org
> > 
> > > Works for me.
> > 
> > Doesn't work for me. Perhaps is Windows-specific.
> 
> Could you please bisect this?

Actually, I think I found the revision to blame.  The problem started
to happen between Dec 25 and Dec 26.  This commit sounds relevant:

  revno: 115757
  committer: martin rudalics <rudalics@gmx.at>
  branch nick: trunk
  timestamp: Thu 2013-12-26 11:31:42 +0000
  message:
    Some more tinkering with Bug#16051.

    * window.c (resize_frame_windows): Don't let the size of the
    root window drop below the frame's default character size.
    Never ever delete any subwindows - let the window manager do the
    clipping.

    * w32fns.c (x_set_tool_bar_lines): Rewrite calculation of number
    of toolbar lines needed when they exceed the height of the root
    window.
    (unwind_create_frame_1): New function.
    (Fx_create_frame): Generally inhibit calling the window
    configuration change hook here.  Remove extra call to
    change_frame_size - it's not needed when we don't run the
    configuration change hook.

Martin, could you take a look, please?

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[martin rudalics]

 > Martin, could you take a look, please?

Should be fixed now (apparently setting face attributes can run Lisp
code from within Fx_create_frame and if that really happens it looks
like a bad idea to me).

martin

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[Eli Zaretskii]

> Date: Sat, 25 Jan 2014 15:59:05 +0100
> From: martin rudalics <rudalics@gmx.at>
> CC: lekktu@gmail.com, 16529@debbugs.gnu.org
> 
>  > Martin, could you take a look, please?
> 
> Should be fixed now

It is for me, thanks.

> (apparently setting face attributes can run Lisp code from within
> Fx_create_frame and if that really happens it looks like a bad idea
> to me).

Why do you think it is a bad idea?

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[martin rudalics]

 >> (apparently setting face attributes can run Lisp code from within
 >> Fx_create_frame and if that really happens it looks like a bad idea
 >> to me).
 >
 > Why do you think it is a bad idea?

Because, as someone remarked, at that time the frame is "not official
yet".  I haven't looked into the details but, if I understand correctly,
NOT running Lisp code in Fx_create_frame can detrimentally affect the
appearance of future frames.  Such design doesn't strike me as
particularly sane.  But maybe all functions run are of the benign type.

martin

bug#16529: 24.3.50; regression: `set-face-attribute' broken for attribute `:font'

[Drew Adams]

> Should be fixed now

Thanks.