GNU bug report logs - #5040
23.1; w32-register-hot-key problem

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 5040 in the body.
You can then email your comments to 5040 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> emacsbugs.donarmstrong.com (full text, mbox):

From: "Drew Adams" <drew.adams <at> oracle.com>
To: <bug-gnu-emacs <at> gnu.org>
Subject: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 13:10:53 -0800

emacs -Q
M-: (w32-register-hot-key [M-tab])
 
Visit a buffer in emacs-lisp-mode.
Type `forward-', then hit `M-TAB'.
 
The Windows TAB dialog comes up (dunno what you call it - a seqence of
icons; you TAB to cycle among their apps) when you hold down the keys
and then disappears when you release them. It should not not come up
at all; instead, `lisp-complete-symbol' should be executed.
 
However, from then on it works as it should. It is only the first time
you press `M-TAB' that Windows doesn't pass the chord to Emacs.
 
emacs -Q
M-: (w32-register-hot-key [M-tab])
 
Visit a buffer in emacs-lisp-mode.  `C-h k M-TAB'. Same as above: the
Windows TAB dialog comes up and then disappears when you release the
keys. And from then on it works OK.
 
In sum, after evalling (w32-register-hot-key [M-tab]) the first
`M-TAB' should be seen by Emacs, but it is not. Only subsequent
`M-TAB's are seen by Emacs.
 
This worked perfectly in Emacs 20 and 21. It is broken starting with
Emacs 22.
 
 
 
 
 
In GNU Emacs 23.1.1 (i386-mingw-nt5.1.2600)
 of 2009-07-29 on SOFT-MJASON
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (4.4)'
 
Important settings:
  value of $LC_ALL: nil
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: ENU
  value of $XMODIFIERS: nil
  locale-coding-system: cp1252
  default-enable-multibyte-characters: t
 
Major mode: Emacs-Lisp
 
Minor modes in effect:
  tooltip-mode: t
  tool-bar-mode: t
  mouse-wheel-mode: t
  menu-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  global-auto-composition-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: t
 
Recent input:
M-: C-y <return> <return> C-h k C-g C-g C-x 1 C-h k 
<M-tab> C-g C-g <help-echo> <help-echo> <help-echo> 
<help-echo> <help-echo> <help-echo> <help-echo> <help-echo> 
<help-echo> <help-echo> <menu-bar> <help-menu> <se
nd-emacs-bug-report>
 
Recent messages:
For information about GNU Emacs and the GNU system, type C-h C-a.
265 (#o411, #x109)
Type C-x 1 to delete the help window.
Quit
Type C-x 1 to delete the help window, C-M-v to scroll help.
Quit
Quit

Message #10 received at submit <at> emacsbugs.donarmstrong.com (full text, mbox):

From: Lennart Borgman <lennart.borgman <at> gmail.com>
To: Drew Adams <drew.adams <at> oracle.com>, 5040 <at> debbugs.gnu.org
Cc: bug-gnu-emacs <at> gnu.org
Subject: Re: bug#5040: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 22:48:49 +0100

On Wed, Nov 25, 2009 at 10:10 PM, Drew Adams <drew.adams <at> oracle.com> wrote:
> emacs -Q
> M-: (w32-register-hot-key [M-tab])
>
> Visit a buffer in emacs-lisp-mode.
> Type `forward-', then hit `M-TAB'.
>
> The Windows TAB dialog comes up (dunno what you call it - a seqence of
> icons; you TAB to cycle among their apps) when you hold down the keys
> and then disappears when you release them. It should not not come up
> at all; instead, `lisp-complete-symbol' should be executed.
>
> However, from then on it works as it should. It is only the first time
> you press `M-TAB' that Windows doesn't pass the chord to Emacs.


This is AFAICS not a bug. M-TAB (and some other keys) can not be
registered as hot keys according to MS documentation.

Message #20 received at submit <at> emacsbugs.donarmstrong.com (full text, mbox):

From: "Drew Adams" <drew.adams <at> oracle.com>
To: "'Lennart Borgman'" <lennart.borgman <at> gmail.com>,
        <5040 <at> debbugs.gnu.org>
Cc: <bug-gnu-emacs <at> gnu.org>
Subject: RE: bug#5040: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 13:52:52 -0800

> On Wed, Nov 25, 2009 at 10:10 PM, Drew Adams 
> <drew.adams <at> oracle.com> wrote:
> > emacs -Q
> > M-: (w32-register-hot-key [M-tab])
> >
> > Visit a buffer in emacs-lisp-mode.
> > Type `forward-', then hit `M-TAB'.
> >
> > The Windows TAB dialog comes up (dunno what you call it - a 
> > seqence of icons; you TAB to cycle among their apps) when
> > you hold down the keys and then disappears when you release
> > them. It should not not come up at all; instead,
> > `lisp-complete-symbol' should be executed.
> >
> > However, from then on it works as it should. It is only the 
> > first time you press `M-TAB' that Windows doesn't pass the
> > chord to Emacs.
> 
> This is AFAICS not a bug. M-TAB (and some other keys) can not be
> registered as hot keys according to MS documentation.

As I said:

1. It works perfectly after the first time you hit `M-TAB'.

2. It works perfectly in Emacs 20 and 21, even the first time.

So AFAICS what you say is simply not true, at least for me (Windows XP SP3).

Message #30 received at 5040 <at> emacsbugs.donarmstrong.com (full text, mbox):

From: Lennart Borgman <lennart.borgman <at> gmail.com>
To: Drew Adams <drew.adams <at> oracle.com>
Cc: 5040 <5040 <at> debbugs.gnu.org>
Subject: Re: bug#5040: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 22:55:04 +0100

> As I said:
>
> 1. It works perfectly after the first time you hit `M-TAB'.
>
> 2. It works perfectly in Emacs 20 and 21, even the first time.
>
> So AFAICS what you say is simply not true, at least for me (Windows XP SP3).


Why does it matter that it worked for you before?

Message #35 received at 5040 <at> emacsbugs.donarmstrong.com (full text, mbox):

From: "Drew Adams" <drew.adams <at> oracle.com>
To: "'Lennart Borgman'" <lennart.borgman <at> gmail.com>
Cc: "'5040'" <5040 <at> debbugs.gnu.org>
Subject: RE: bug#5040: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 14:01:46 -0800

> > As I said:
> > 1. It works perfectly after the first time you hit `M-TAB'.
> > 2. It works perfectly in Emacs 20 and 21, even the first time.
> >
> > So AFAICS what you say is simply not true, at least for me 
> > (Windows XP SP3).
> 
> Why does it matter that it worked for you before?

Huh? Who said anything about "before"?

It works for me NOW. And probably for others too.
It works in Emacs 20 and 21 always,
   including the first time I hit `M-TAB'.
It works in Emacs 22 and 23 also,
   EXCEPT for the first time I hit `M-TAB'.

This is a regression starting in Emacs 22. We know it is possible for it to work
starting with the first `M-TAB', because it does (still) work in older Emacs
versions (now, not "before').

Message #40 received at 5040 <at> emacsbugs.donarmstrong.com (full text, mbox):

From: Lennart Borgman <lennart.borgman <at> gmail.com>
To: Drew Adams <drew.adams <at> oracle.com>
Cc: 5040 <5040 <at> debbugs.gnu.org>
Subject: Re: bug#5040: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 23:04:24 +0100

On Wed, Nov 25, 2009 at 11:01 PM, Drew Adams <drew.adams <at> oracle.com> wrote:
>> > As I said:
>> > 1. It works perfectly after the first time you hit `M-TAB'.
>> > 2. It works perfectly in Emacs 20 and 21, even the first time.
>> >
>> > So AFAICS what you say is simply not true, at least for me
>> > (Windows XP SP3).
>>
>> Why does it matter that it worked for you before?
>
> Huh? Who said anything about "before"?
>
> It works for me NOW. And probably for others too.
> It works in Emacs 20 and 21 always,
>   including the first time I hit `M-TAB'.
> It works in Emacs 22 and 23 also,
>   EXCEPT for the first time I hit `M-TAB'.
>
> This is a regression starting in Emacs 22. We know it is possible for it to work
> starting with the first `M-TAB', because it does (still) work in older Emacs
> versions (now, not "before').


No, it is not guaranteed to work. If you want to be sure that it works
you need something like the low level keyboard path I have suggested
before.

Message #45 received at 5040 <at> emacsbugs.donarmstrong.com (full text, mbox):

From: "Drew Adams" <drew.adams <at> oracle.com>
To: "'Lennart Borgman'" <lennart.borgman <at> gmail.com>
Cc: "'5040'" <5040 <at> debbugs.gnu.org>
Subject: RE: bug#5040: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 14:18:03 -0800

> >> > As I said:
> >> > 1. It works perfectly after the first time you hit `M-TAB'.
> >> > 2. It works perfectly in Emacs 20 and 21, even the first time.
> >> >
> >> > So AFAICS what you say is simply not true, at least for me
> >> > (Windows XP SP3).
> >>
> >> Why does it matter that it worked for you before?
> >
> > Huh? Who said anything about "before"?
> >
> > It works for me NOW. And probably for others too.
> > It works in Emacs 20 and 21 always,
> >   including the first time I hit `M-TAB'.
> > It works in Emacs 22 and 23 also,
> >   EXCEPT for the first time I hit `M-TAB'.
> >
> > This is a regression starting in Emacs 22. We know it is 
> > possible for it to work starting with the first `M-TAB',
> > because it does (still) work in older Emacs versions (now,
> > not "before').
> 
> > No, it is not guaranteed to work. If you want to be sure that it works
> > you need something like the low level keyboard path I have suggested
> > before.

Lennart, we don't seem to be communicating.
"No", what?  Did I say anything about guarantees?

I said it works FOR ME (Windows XP SP3), and I suspect for at least some others
also. (Have you tried it (with emacs -Q)? Does it work for you? You might be
surprised that it does.)

What is THIS bug report about? There is a minor regression starting with Emacs
22. THAT'S ALL.

Please don't turn this into a discussion/rant about the Windows/Emacs patching
you "have suggested before". All I'm asking for is that this particular bug be
fixed - to make Emacs do again what it did in Emacs 21. Nothing more.

Let's not let the perfect become the enemy of the good. If you want to file
another bug about guarantees, low-level patches, and other things, feel free. I
have no objection to perfection ;-). But while waiting, let's see if this can be
fixed to at least let Emacs do what it did before.

Message #50 received at 5040 <at> emacsbugs.donarmstrong.com (full text, mbox):

From: Lennart Borgman <lennart.borgman <at> gmail.com>
To: Drew Adams <drew.adams <at> oracle.com>
Cc: 5040 <5040 <at> debbugs.gnu.org>
Subject: Re: bug#5040: 23.1; w32-register-hot-key problem
Date: Wed, 25 Nov 2009 23:24:17 +0100

On Wed, Nov 25, 2009 at 11:18 PM, Drew Adams <drew.adams <at> oracle.com> wrote:
>
> Lennart, we don't seem to be communicating.
> "No", what?  Did I say anything about guarantees?
>
> I said it works FOR ME (Windows XP SP3), and I suspect for at least some others
> also. (Have you tried it (with emacs -Q)? Does it work for you? You might be
> surprised that it does.)
>
> What is THIS bug report about? There is a minor regression starting with Emacs
> 22. THAT'S ALL.

I am trying to say that this is not guaranteed to work. Some key
combinations are strictly not allowed by w32 to be used by
applications.

And I think I misremembered. I am not sure if it can work with a low
level keyboard hook either.

There is not very much that we can do to guarantee that it works. If
it works it works a bit by chance. Next patch from MS may stop it from
working.

But you are right, if it works for you with the old versions of Emacs
there might be a way to fool w32 here.

Message #57 received at 5040-done <at> debbugs.gnu.org (full text, mbox):

From: Juanma Barranquero <lekktu <at> gmail.com>
To: Lennart Borgman <lennart.borgman <at> gmail.com>,
	5040-done <at> emacsbugs.donarmstrong.com
Cc: Drew Adams <drew.adams <at> oracle.com>
Subject: Re: bug#5040: 23.1; w32-register-hot-key problem
Date: Sat, 2 Jul 2011 04:04:03 +0200

On one hand, Drew is right that if it works with 21.X and 22.X, it
should work with 23.X+.

On the other hand, Lennart's right that Alt-Tab is reserved and, to
make it work reliably it would be necessary to write a low level
keyboard hook.

On the third hand, alt-tab role in Vista and Windows 7 is enhanced and
it does not make much sense to try to bypass it and go against the
system (not that it does not make any sense, just not worth the effort
IMHO). So I'm closing this one.

    Juanma

Message #58 received at 5040-done <at> debbugs.gnu.org (full text, mbox):

From: "Drew Adams" <drew.adams <at> oracle.com>
To: "'Juanma Barranquero'" <lekktu <at> gmail.com>,
	"'Lennart Borgman'" <lennart.borgman <at> gmail.com>,
	<5040-done <at> emacsbugs.donarmstrong.com>
Subject: RE: bug#5040: 23.1; w32-register-hot-key problem
Date: Sat, 2 Jul 2011 07:16:31 -0700

> On one hand, Drew is right that if it works with 21.X and 22.X, it
> should work with 23.X+.

Correction: it works with 20 and 21, not with 22 or 23 (or 24).

And it works in all versions after just the first M-TAB.  It is only the first
M-TAB in Emacs 22+ where it does not work.  Why not at least look into why this
might be?

Emacs apparently *knows* how to make this work - it works forever after the
first M-TAB, and in Emacs 20 and 21 it works even for the first M-TAB.  It's not
like this is something Emacs doesn't know how to do correctly - it's that a bug
was introduced to break it.

> On the other hand, Lennart's right that Alt-Tab is reserved and, to
> make it work reliably it would be necessary to write a low level
> keyboard hook.
> 
> On the third hand, alt-tab role in Vista and Windows 7 is enhanced and
> it does not make much sense to try to bypass it and go against the
> system (not that it does not make any sense, just not worth the effort
> IMHO). So I'm closing this one.

I don't know anything about Vista or Windows 7, so I can't speak to that.

That MS Windows "reserves" something is not in itself (i.e., alone) a reason
that Emacs should respect that "reservation".  There is nothing new about MS
Windows using M-TAB, and nothing new about Emacs using M-TAB.

And to use it in Emacs, Windows users must explicitly do (w32-register-hot-key
[M-tab]).  That's a user choice, not Emacs's choice or Windows's choice.

The bug remains.  And AFAICT no one has even looked at it (why does it happen?
what change introduced the bug?).  There has only been argument over whether
Windows officially permits Emacs to do something it has always done.  There is
nothing in this bug thread apart from the bug report recipe and frothing about
how Emacs shouldn't disrespect Windows's "reservation".

Why the zeal to "close" bugs that remain bugs?  This bug has never even been
investigated.  If there is insufficient manpower to look into this bug, then
leave it open.  Or if Emacs Dev is certain that it should not fix it (why?),
then change it to `wontfix'.  It is a bug nonetheless - and a regression no
less.

Message #59 received at 5040-done <at> debbugs.gnu.org (full text, mbox):

From: Lennart Borgman <lennart.borgman <at> gmail.com>
To: Drew Adams <drew.adams <at> oracle.com>
Cc: Juanma Barranquero <lekktu <at> gmail.com>, 5040-done <at> emacsbugs.donarmstrong.com
Subject: Re: bug#5040: 23.1; w32-register-hot-key problem
Date: Sat, 2 Jul 2011 16:37:56 +0200

On Sat, Jul 2, 2011 at 16:16, Drew Adams <drew.adams <at> oracle.com> wrote:
>
> That MS Windows "reserves" something is not in itself (i.e., alone) a reason
> that Emacs should respect that "reservation".  There is nothing new about MS
> Windows using M-TAB, and nothing new about Emacs using M-TAB.
>
> And to use it in Emacs, Windows users must explicitly do (w32-register-hot-key
> [M-tab]).  That's a user choice, not Emacs's choice or Windows's choice.

MS has warned not to use this combination and some other. It might be
possible to sometimes get around this technically but that is not
guaranteed.

As I think I have told in a bug report I have seen more problems in
this area in Windows 7 than before.

In addition to this I have seen some very strange things in this area
in Windows 7 (64-bit) when running 32-bit Emacs. However I suspect
that might be problems with gcc compiler. (And the lack of control of
status after system calls on w32 in Emacs. Something that makes it
hard to catch problems of the type I have seen now. At the moment I
have not time to look at it. Sadly I do not have time to argue about
adding these checks either, but I would be glad if they were added.)

Message #60 received at 5040-done <at> debbugs.gnu.org (full text, mbox):

From: Juanma Barranquero <lekktu <at> gmail.com>
To: Drew Adams <drew.adams <at> oracle.com>
Cc: 5040-done <at> emacsbugs.donarmstrong.com,
	Lennart Borgman <lennart.borgman <at> gmail.com>
Subject: Re: bug#5040: 23.1; w32-register-hot-key problem
Date: Sat, 2 Jul 2011 19:23:47 +0200

On Sat, Jul 2, 2011 at 16:16, Drew Adams <drew.adams <at> oracle.com> wrote:

> Correction: it works with 20 and 21, not with 22 or 23 (or 24).

Yes, sorry.

> Why not at least look into why this
> might be?
>
> Emacs apparently *knows* how to make this work - it works forever after the
> first M-TAB, and in Emacs 20 and 21 it works even for the first M-TAB.  It's not
> like this is something Emacs doesn't know how to do correctly - it's that a bug
> was introduced to break it.

Or perhaps it worked right before just by happenstance. Perhaps we
were passing uninitialized data to some API, and the moment we fixed
it it stopped working. On Windows 7, I cannot make it work *at all*,
not the first time, not afterwards, not ever.

> That MS Windows "reserves" something is not in itself (i.e., alone) a reason
> that Emacs should respect that "reservation".  There is nothing new about MS
> Windows using M-TAB, and nothing new about Emacs using M-TAB.
>
> And to use it in Emacs, Windows users must explicitly do (w32-register-hot-key
> [M-tab]).  That's a user choice, not Emacs's choice or Windows's choice.
>
> The bug remains.  And AFAICT no one has even looked at it (why does it happen?
> what change introduced the bug?).  There has only been argument over whether
> Windows officially permits Emacs to do something it has always done.  There is
> nothing in this bug thread apart from the bug report recipe and frothing about
> how Emacs shouldn't disrespect Windows's "reservation".

> Why the zeal to "close" bugs that remain bugs?  This bug has never even been
> investigated.  If there is insufficient manpower to look into this bug, then
> leave it open.

There are very few maintainers of Emacs for Windows. No one of us has
shown any inclination to fix it or give it too much weight, and even
if you believe that Emacs is not bound to obey Microsoft, it is
certainly easier to go with the flow that against it, and the flow is
for that kind of keystrokes to be more strongly reserved for Windows
functions, not less.

It *was* a regression, on 2007-06-02 when 22.1 was released. Now, four
years later, it seems like something which is not worth investing any
effort in fixing it (specially because, as I've said, I think it
wouldn't easily work at all on newer Windows). Of course that's my
take, and that's why I closed the bug. If you feel strongly, don't
hesitate to reopen it and change it to a Wontfix, because I suspect no
one is going to go the trouble to fix it anyway (though I could be
wrong).

    Juanma

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.