NTEmacs problems with make-frame

NTEmacs problems with make-frame

[Eli Zaretskii]

For quite some time now I sometimes see an annoying problem in the
native MS-Windows (a.k.a. NTEmacs) port of the CVS Emacs: commands
that create a new frame somehow fail to display the frame.  Examples
include "C-x b FOO RET", "C-x m" with rmail-mail-new-frame set to
non-nil, etc.  What happens is that the frame is created (it appears
in the output of `frame-list'), it is not (and cannot be) displayed.
Instead, the frame that was current becomes unselected (the cursor
changes into a hollow block), and that's it.  Emacs is not stuck: if I
move the mouse out of the Emacs frame and back into it, the frame that
was selected before the offending command becomes again selected and
everything is hunky-dory (except that there's a frame that I cannot
switch to).

This problem happens in about 50% of the attempts to use the commands
that create a new frame.

I've seen this on two different machines running Windows XP, and in
several versions since last October, up to and including the latest
binary I have, which was compiled about a month ago (I didn't yet have
time to set up a MinGW build environment, so I rely on binaries built
by others).

Does anyone else see this on MS-Windows?  The only thing that is
both peculiar and common to the two machines where I have this problem
is that both of them have Active Window Tracking (a.k.a. focus follows
mouse) feature of the window manager set to ON.

Suggestions for further testing and requests for additional
information are welcome.

TIA

Re: NTEmacs problems with make-frame

[Eli Zaretskii]

> Date: Wed, 25 May 2005 21:49:47 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> 
> For quite some time now I sometimes see an annoying problem in the
> native MS-Windows (a.k.a. NTEmacs) port of the CVS Emacs: commands
> that create a new frame somehow fail to display the frame.  Examples
> include "C-x b FOO RET", "C-x m" with rmail-mail-new-frame set to
           ^^^^^^^^^^^^^
I meant "C-x 5 b FOO RET", of course.

Re: NTEmacs problems with make-frame

[Juanma Barranquero]

> native MS-Windows (a.k.a. NTEmacs) port of the CVS Emacs: commands
> that create a new frame somehow fail to display the frame.

I doesn't happen for my, on XP, with neither MSVC nor MinGW builds.
I've tested *without* Active Window Tracking. I'll enable it next time
I reboot Windows to see whether that has any effect.

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

Re: NTEmacs problems with make-frame

[Neil Jerram]

Eli Zaretskii wrote:
> For quite some time now I sometimes see an annoying problem in the
> native MS-Windows (a.k.a. NTEmacs) port of the CVS Emacs: commands
> that create a new frame somehow fail to display the frame.  [...]

Yes, this happens for me too on Windows 2000.  With Ediff, for example,
sometimes this happens with the Ediff control frame.

> Does anyone else see this on MS-Windows?  The only thing that is
> both peculiar and common to the two machines where I have this problem
> is that both of them have Active Window Tracking (a.k.a. focus follows
> mouse) feature of the window manager set to ON.

I don't have this.  I have out-of-the-box W2K behaviour, which is click
to focus.

	Neil