GNU bug report logs - #17090
desktop.el saves unusable register data in register-alist

Previous Next

To reply to this bug, email your comments to 17090 AT debbugs.gnu.org.

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

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Juanma Barranquero <lekktu <at> gmail.com>
To: Bug-Gnu-Emacs <bug-gnu-emacs <at> gnu.org>
Subject: desktop.el saves unusable register data in register-alist
Date: Mon, 24 Mar 2014 21:31:30 +0100

Package: Emacs
Severity: minor

emacs -Q -f desktop-save-mode
C-x r w a
C-x C-x
y <RET>
emacs -Q -f desktop-read
C-x r j

=> a: rectangle starting with Unprintable entity

and similarly, when using
C-x r f a

restoring and then C-x r j  brings

Error running timer: (invalid function "Unprintable entity")

register-alist is in desktop-globals-to-save by default, so it should
work out of the box. but any fix is going to be ugly.

Possibilities:

- Cleaning register-alist from desktop-save-hook. Not a good idea
because the session does not necessarily end on desktop-saving (and
even less now, with timer-based saving).

- Modify desktop-outvar to skip anything whose serialization matches
"Unprintable entity".

- Modify desktop-outvar with register-alist specific knowledge, so it
filters out only some entries when serializing.

- Design a more generic mechanism to allow variables to have cleaning
specific functions.

The first three options are horribly hackish, and the last one is a
bit overkill if we only want to avoid the register-alist issue (the
mechanism would be generally useful, but no one has ever asked for
it...)

Ideas?

Message #8 received at 17090 <at> debbugs.gnu.org (full text, mbox):

From: Lars Ingebrigtsen <larsi <at> gnus.org>
To: Juanma Barranquero <lekktu <at> gmail.com>
Cc: 17090 <at> debbugs.gnu.org
Subject: Re: bug#17090: desktop.el saves unusable register data in
 register-alist
Date: Sun, 29 Sep 2019 16:42:32 +0200

Juanma Barranquero <lekktu <at> gmail.com> writes:

> emacs -Q -f desktop-save-mode
> C-x r w a
> C-x C-x
> y <RET>
> emacs -Q -f desktop-read
> C-x r j
>
> => a: rectangle starting with Unprintable entity
>
> and similarly, when using
> C-x r f a
>
> restoring and then C-x r j  brings
>
> Error running timer: (invalid function "Unprintable entity")

I'm unable to reproduce this bug in Emacs 27 (and I also vaguely
remember something being fixed in this area a few months ago).

Instead of the error, I get:

Desktop: 1 frame, 0 buffers restored.

and then 

Register doesn’t contain a buffer position or configuration

when trying to C-x r j a-ing.

Are you still seeing this problem in Emacs 27?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Message #15 received at 17090 <at> debbugs.gnu.org (full text, mbox):

From: Lars Ingebrigtsen <larsi <at> gnus.org>
To: Juanma Barranquero <lekktu <at> gmail.com>
Cc: 17090 <at> debbugs.gnu.org
Subject: Re: bug#17090: desktop.el saves unusable register data in
 register-alist
Date: Sun, 29 Sep 2019 16:59:59 +0200

Lars Ingebrigtsen <larsi <at> gnus.org> writes:

> I'm unable to reproduce this bug in Emacs 27 (and I also vaguely
> remember something being fixed in this area a few months ago).

Hm, I think I was wrong here -- I'm not getting exactly the same errors
as in earlier versions of Emacs, but the data that's saved still isn't
usable.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Message #20 received at 17090 <at> debbugs.gnu.org (full text, mbox):

From: Juanma Barranquero <lekktu <at> gmail.com>
To: Lars Ingebrigtsen <larsi <at> gnus.org>
Cc: 17090 <at> debbugs.gnu.org
Subject: Re: bug#17090: desktop.el saves unusable register data in
 register-alist
Date: Sun, 29 Sep 2019 18:17:24 +0200

[Message part 1 (text/plain, inline)]

Curious that my original recipe (fixing the C-x C-x to C-x C-c, as I
suppose I intended) doesn't work now because

  emacs -Q -f desktop-read

does not read the desktop file from the current directory ('.' is no longer
in desktop-path).

But anyway, back to the issue:

> Hm, I think I was wrong here -- I'm not getting exactly the same errors
> as in earlier versions of Emacs, but the data that's saved still isn't
> usable.

Correct.

But, I suppose it is perhaps not worth fixing now, because after that we
implemented framesets, and you can do

C-x r f a

and the frameset in register a is saved and correctly reloaded afterwards.
It's a bit of an overkill, of course.

If we want to fix it, perhaps we could trick our way, by using framesets to
save a one-frame frameset with some special info to mark that it is really
intended to be just a window-configuration, and then, when reloading the
desktop, restoring somehow the window configuration into the register, from
the frameset data. No idea if it is worth the effort, but I think it would
be doable.

[Message part 2 (text/html, inline)]

Message #23 received at 17090 <at> debbugs.gnu.org (full text, mbox):

From: Lars Ingebrigtsen <larsi <at> gnus.org>
To: Juanma Barranquero <lekktu <at> gmail.com>
Cc: 17090 <at> debbugs.gnu.org
Subject: Re: bug#17090: desktop.el saves unusable register data in
 register-alist
Date: Mon, 30 Sep 2019 06:40:19 +0200

Juanma Barranquero <lekktu <at> gmail.com> writes:

> But, I suppose it is perhaps not worth fixing now, because after that
> we implemented framesets, and you can do
>
> C-x r f a
>
> and the frameset in register a is saved and correctly reloaded
> afterwards. It's a bit of an overkill, of course.
>
> If we want to fix it, perhaps we could trick our way, by using
> framesets to save a one-frame frameset with some special info to mark
> that it is really intended to be just a window-configuration, and
> then, when reloading the desktop, restoring somehow the window
> configuration into the register, from the frameset data. No idea if it
> is worth the effort, but I think it would be doable.

It sounds like we should perhaps refer users to the C-x r f a command
instead...  But in any case, desktop.el probably shouldn't be saving
these useless window configuration registers.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Message #26 received at 17090 <at> debbugs.gnu.org (full text, mbox):

From: Juanma Barranquero <lekktu <at> gmail.com>
To: Lars Ingebrigtsen <larsi <at> gnus.org>
Cc: 17090 <at> debbugs.gnu.org
Subject: Re: bug#17090: desktop.el saves unusable register data in
 register-alist
Date: Mon, 30 Sep 2019 10:10:30 +0200

[Message part 1 (text/plain, inline)]

On Mon, Sep 30, 2019 at 6:40 AM Lars Ingebrigtsen <larsi <at> gnus.org> wrote:

> It sounds like we should perhaps refer users to the C-x r f a command
> instead...

Currently, the function assigned to the C-x r map to save a frameset saves
all the frames, so in some uses, saving and restoring the window
configuration (which affects just one frame) is faster and cleaner. Also,
restoring a frameset not only restores the window configuration, but also
position on the screen.

Of course, it would be possible to add another command, or a prefix
parameter, to frameset-to-register, to save just the selected frame, and to
restore the frameset without moving the frame.

> But in any case, desktop.el probably shouldn't be saving
> these useless window configuration registers.

Definitely. It's wasted space and time, and does not serve any useful
purpose.

FWIW, I agree. Documenting the issue and filtering out the
window-configurations on desktop save seems like the easier answer right
now. Don't think it's really a very urgent or frequent issue.

[Message part 2 (text/html, inline)]

Message #29 received at 17090 <at> debbugs.gnu.org (full text, mbox):

From: Juri Linkov <juri <at> linkov.net>
To: Juanma Barranquero <lekktu <at> gmail.com>
Cc: 17090 <at> debbugs.gnu.org, Lars Ingebrigtsen <larsi <at> gnus.org>
Subject: Re: bug#17090: desktop.el saves unusable register data in
 register-alist
Date: Mon, 07 Oct 2019 21:28:50 +0300

> If we want to fix it, perhaps we could trick our way, by using framesets to
> save a one-frame frameset with some special info to mark that it is really
> intended to be just a window-configuration, and then, when reloading the
> desktop, restoring somehow the window configuration into the register, from
> the frameset data. No idea if it is worth the effort, but I think it would
> be doable.

I invite you to look at the solution that Martin helped to develop
that restores tabs with window-configurations from the desktop file
using framesets, see the comment in tab-bar.el (just search
for “window-configuration”).

Previous Next

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