22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Martin Fischer]

Please write in English if possible, because the Emacs maintainers
usually do not have translators to read other languages for them.

Your bug report will be posted to the emacs-pretest-bug@gnu.org mailing list.

Please describe exactly what actions triggered the bug
and the precise symptoms of the bug:

  8<--------------------------------------------------------------------

  ...\emacs-22.1.90\bin\runemacs.exe -Q

  Customize the variable uniquify-buffer-name-style to "reverse".

  List the directories c:\otto\ludwig and c:\karl\ludwig with dired.

  M-x toggle-debug-on-error RET

  M-x desktop-save RET

    Debugger entered--Lisp error: (error "No buffer named ludwig")
      desktop-save-buffer-p(nil "ludwig" dired-mode nil 185 (nil nil) t ("c:/otto/ludwig/") nil)
      apply(desktop-save-buffer-p (nil "ludwig" dired-mode nil 185 (nil nil) t ("c:/otto/ludwig/") nil))
      desktop-save("~/")
      call-interactively(desktop-save)
      execute-extended-command(nil)
      call-interactively(execute-extended-command)

  M-x list-buffers RET

    .%* *Backtrace*            392  Debugger
     %  *Completions*          167  Completion List
     %  ludwig\karl\           184  Dired by name	  c:/karl/ludwig/
     %  ludwig\otto\           184  Dired by name	  c:/otto/ludwig/
        *scratch*                0  Lisp Interaction
      * *Messages*             957  Fundamental

  8<--------------------------------------------------------------------

If Emacs crashed, and you have the Emacs process in the gdb debugger,
please include the output from the following gdb commands:
    `bt full' and `xbacktrace'.
If you would like to further debug the crash, please read the file
d:/aemf/emacs-22.1.90/etc/DEBUG for instructions.


In GNU Emacs 22.1.90.1 (i386-mingw-nt5.1.2600)
 of 2008-01-31 on PRETEST
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4) --cflags -Ic:/gnuwin32/include'

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: DEU
  locale-coding-system: cp1252
  default-enable-multibyte-characters: t

Major mode: Debugger

Minor modes in effect:
  shell-dirtrack-mode: t
  encoded-kbd-mode: t
  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
  unify-8859-on-encoding-mode: t
  utf-translate-cjk-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: identity

Recent input:
<escape> x c u s t <tab> - g <tab> <return> u n i q 
<tab> <return> <help-echo> <down-mouse-1> <mouse-1> 
<help-echo> <down-mouse-1> <mouse-1> <help-echo> <down-mouse-1> 
<help-echo> <help-echo> <help-echo> C-x d c : / <return> 
+ t e s t - 1 / t e s t <backspace> <backspace> <backspace> 
<backspace> <backspace> <backspace> <backspace> <backspace> 
<backspace> <backspace> <backspace> o t o <backspace> 
t o / l u d w i g <return> + <up> <backspace> <backspace> 
<backspace> <backspace> <backspace> <backspace> <backspace> 
<return> + <up> <up> <return> <up> + <up> <up> <backspace> 
<backspace> <backspace> <backspace> k a r l <return> 
+ <up> <up> <left> <left> <left> <left> <left> <left> 
<left> <backspace> <backspace> <backspace> <backspace> 
k a r l <return> C-x d k a r l <tab> l <tab> <return> 
C-x d o t t <tab> <backspace> <backspace> <backspace> 
<backspace> <backspace> <backspace> <backspace> <backspace> 
<backspace> <backspace> <backspace> <backspace> <backspace> 
<backspace> <backspace> o t <tab> l u <tab> <return> 
<escape> x d e s k t o <tab> s <tab> <return> C-g C-g 
<escape> x t o g g <tab> d e b <tab> e <tab> <return> 
<escape> x d e s k t o p <tab> s a v <tab> <return> 
~ / <tab> <backspace> <backspace> <backspace> <backspace> 
<backspace> <backspace> <backspace> <backspace> <backspace> 
<return> <up> <lwindow> <help-echo> <help-echo> <help-echo> 
<help-echo> <help-echo> <help-echo> <help-echo> <help-echo> 
<help-echo> <help-echo> <help-echo> <menu-bar> <help-menu> 
<send-emacs-bug-report>

Recent messages:
Loading desktop...done
Quit [2 times]
Debug on Error enabled globally
Wrote c:/Dokumente und Einstellungen/aemf/home-empty-06.02.2008/.emacs.desktop.lock
Loading debug...done
Entering debugger...
Loading emacsbug...done
call-interactively: Text is read-only
next-history-element: Beginning of history; no preceding item
byte-code: Buffer is read-only: #<buffer *Backtrace*>

-- 
parozusa at web dot de

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Feb 6, 2008 5:12 PM, Martin Fischer <parozusa@web.de> wrote:

>   Customize the variable uniquify-buffer-name-style to "reverse".

>     Debugger entered--Lisp error: (error "No buffer named ludwig")
>       desktop-save-buffer-p(nil "ludwig" dired-mode nil 185 (nil nil) t ("c:/otto/ludwig/") nil)
>       apply(desktop-save-buffer-p (nil "ludwig" dired-mode nil 185 (nil nil) t ("c:/otto/ludwig/") nil))
>       desktop-save("~/")
>       call-interactively(desktop-save)
>       execute-extended-command(nil)
>       call-interactively(execute-extended-command)

That's related to a change I did a while ago. I'll take a look.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Feb 6, 2008 5:12 PM, Martin Fischer <parozusa@web.de> wrote:

>     Debugger entered--Lisp error: (error "No buffer named ludwig")
>       desktop-save-buffer-p(nil "ludwig" dired-mode nil 185 (nil nil) t ("c:/otto/ludwig/") nil)
>       apply(desktop-save-buffer-p (nil "ludwig" dired-mode nil 185 (nil nil) t ("c:/otto/ludwig/") nil))
>       desktop-save("~/")
>       call-interactively(desktop-save)
>       execute-extended-command(nil)
>       call-interactively(execute-extended-command)

Plese try the following patch. Is a bit of hackish, but doing it
cleanly would need more changes to desktop.el and perhaps changing the
desktop file version, which is a no-no during the pretest.

             Juanma


Index: lisp/desktop.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/desktop.el,v
retrieving revision 1.108.2.6
diff -u -2 -r1.108.2.6 desktop.el
--- lisp/desktop.el	7 Jan 2008 02:44:57 -0000	1.108.2.6
+++ lisp/desktop.el	8 Feb 2008 16:46:43 -0000
@@ -654,9 +654,10 @@
   (set-buffer buffer)
   (list
+   ;; base name of the buffer; replaces the buffer name if managed by uniquify
+   (and (bound-and-true-p uniquify-managed)
+	(uniquify-item-base (car uniquify-managed)))
    ;; basic information
    (desktop-file-name (buffer-file-name) desktop-dirname)
-   (if (bound-and-true-p uniquify-managed)
-       (uniquify-item-base (car uniquify-managed))
-     (buffer-name))
+   (buffer-name)
    major-mode
    ;; minor modes
@@ -887,5 +888,5 @@
 	  (insert "\n;; Buffer section -- buffers listed in same order as in
buffer list:\n")
 	  (dolist (l (mapcar 'desktop-buffer-info (buffer-list)))
-	    (when (apply 'desktop-save-buffer-p l)
+	    (when (apply 'desktop-save-buffer-p (cdr l))
 	      (insert "("
 		      (if (or (not (integerp eager))
@@ -897,4 +898,6 @@
 		      " "
 		      desktop-file-version)
+	      (let ((base (pop l)))
+		(when base (setcar (nthcdr 1 l) base)))
 	      (dolist (e l)
 		(insert "\n  " (desktop-value-to-string e)))

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Stefan Monnier]

> +   ;; base name of the buffer; replaces the buffer name if managed by uniquify
> +   (and (bound-and-true-p uniquify-managed)
> +	(uniquify-item-base (car uniquify-managed)))
>     ;; basic information
>     (desktop-file-name (buffer-file-name) desktop-dirname)
> -   (if (bound-and-true-p uniquify-managed)
> -       (uniquify-item-base (car uniquify-managed))
> -     (buffer-name))
> +   (buffer-name)
>     major-mode
>     ;; minor modes

Yuck!
Could you explain what is the problem, so we can try and find
a cleaner workaround?


        Stefan

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

> Yuck!

Well, yeah, but let's not exaggerate...

> Could you explain what is the problem, so we can try and find
> a cleaner workaround?

The original problem was that desktop.el saved the uniquified names,
so afterwards (in subsequent invocations of Emacs) the names were
"uniquified" but not really managed by uniquify. You started with A/C
and B/C, saved, reloaded, deleted A/C and still had B/C (instead of
just C).

The fix is simple: save the base name of the buffer instead of the
uniquified one.

That was done by modifying the information returned by
`desktop-buffer-info', which is the information written to the desktop
file for each buffer.

That works fine, except that dired buffers are treated specially,
which triggers the OP problem.

When trying to decide whether a buffer has to be saved or not,
`desktop-save' passes the output of `desktop-buffer-info' to
`desktop-save-buffer-p', which in most cases only takes into account
the name of the visited file. For dired buffers, though, it also looks
at its second argument (the buffer name) to find the buffer. With my
previous change,that fails because the "buffer name" it gets is now
the base name, not the name of the real buffer (that was my mistake, I
didn't realize the special treatment of dired buffers).

Most fixes involve modifying the output of `desktop-buffer-info' to
return both the buffer-name and the base name; however, the list has
to be "fixed back" before saving it to the desktop. Or, alternatively,
you can opt *not* to modify the info list, and instead try to detect
whether the buffer is uniquify-managed at the point you're writing the
info list to the desktop; but that is also unclean because at that
point you shouldn't need to process any element of the list specially,
or know about its contents.

The cleanest fix would be to admit that the info list should have both
the buffer name and the base name. But that means bumping up the
version of the desktop file.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Stefan Monnier]

> The original problem was that desktop.el saved the uniquified names,
> so afterwards (in subsequent invocations of Emacs) the names were
> "uniquified" but not really managed by uniquify. You started with A/C
> and B/C, saved, reloaded, deleted A/C and still had B/C (instead of
> just C).

Hmm...

> The cleanest fix would be to admit that the info list should have both
> the buffer name and the base name. But that means bumping up the
> version of the desktop file.

Actually, any use of the "base name" will be uniquify-specific as well,
so I'll still say "Yuck!".

But I guess it's difficult to avoid it.  But instead of using uniquify
internal data such as uniquify-managed, please change uniquify.el so it
provides the necessary function, e.g. `uniquify-base-name', with an
explanation of why/where it is used/needed.

Indeed for uniquified buffers, you should only [need to] save the
base-name to disk, so there's no need to change the disk representation
of your data.


        Stefan

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Feb 8, 2008 10:17 PM, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> Actually, any use of the "base name" will be uniquify-specific as well,
> so I'll still say "Yuck!".

Well, I'm trying to fix a problem derived of the interaction of
uniquify and desktop, so using uniquify-specific info seems
unavoidable.

> Indeed for uniquified buffers, you should only [need to] save the
> base-name to disk, so there's no need to change the disk representation
> of your data.

The "base name" and the buffer name are two different pieces of info;
even if no uniquify is present, it would be perhaps interesting to
know (and save) whether "my-buffer<2>" is called so by user fiat or
via rename-buffer's UNIQUE argument; but that is not needed right now.

Anyway. Does this version lessen your yucky factor somewhat?

             Juanma


Index: lisp/uniquify.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/uniquify.el,v
retrieving revision 1.62.2.7
diff -u -2 -r1.62.2.7 uniquify.el
--- lisp/uniquify.el	7 Jan 2008 02:44:10 -0000	1.62.2.7
+++ lisp/uniquify.el	12 Feb 2008 00:10:32 -0000
@@ -190,4 +190,10 @@
 (put 'uniquify-managed 'permanent-local t)

+;; Used in desktop.el to save the non-uniquified buffer name
+(defun uniquify-buffer-base-name ()
+  "Return the base name of the current buffer."
+  (and (bound-and-true-p uniquify-managed)
+       (uniquify-item-base (car uniquify-managed))))
+
 ;;; Main entry point.

Index: lisp/desktop.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/desktop.el,v
retrieving revision 1.108.2.6
diff -u -2 -r1.108.2.6 desktop.el
--- lisp/desktop.el	7 Jan 2008 02:44:57 -0000	1.108.2.6
+++ lisp/desktop.el	12 Feb 2008 00:14:02 -0000
@@ -136,6 +136,4 @@
 ;;; Code:

-(defvar uniquify-managed)
-
 (defvar desktop-file-version "206"
   "Version number of desktop file format.
@@ -654,9 +652,9 @@
   (set-buffer buffer)
   (list
+   ;; base name of the buffer; replaces the buffer name if managed by uniquify
+   (and (fboundp 'uniquify-buffer-base-name) (uniquify-buffer-base-name))
    ;; basic information
    (desktop-file-name (buffer-file-name) desktop-dirname)
-   (if (bound-and-true-p uniquify-managed)
-       (uniquify-item-base (car uniquify-managed))
-     (buffer-name))
+   (buffer-name)
    major-mode
    ;; minor modes
@@ -887,5 +885,5 @@
 	  (insert "\n;; Buffer section -- buffers listed in same order as in
buffer list:\n")
 	  (dolist (l (mapcar 'desktop-buffer-info (buffer-list)))
-	    (when (apply 'desktop-save-buffer-p l)
+	    (when (apply 'desktop-save-buffer-p (cdr l))
 	      (insert "("
 		      (if (or (not (integerp eager))
@@ -897,4 +895,7 @@
 		      " "
 		      desktop-file-version)
+	      (let ((base (pop l)))
+		;; If the base name is non-nil, we substitute it for the buffer name
+		(when base (setcar (nthcdr 1 l) base)))
 	      (dolist (e l)
 		(insert "\n  " (desktop-value-to-string e)))

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Stefan Monnier]

>> Actually, any use of the "base name" will be uniquify-specific as well,
>> so I'll still say "Yuck!".

> Well, I'm trying to fix a problem derived of the interaction of
> uniquify and desktop, so using uniquify-specific info seems
> unavoidable.

Yes, that's what I figured as well, which is why I think it's fine to do
it, but only if the interface between the two is made cleaner.

>> Indeed for uniquified buffers, you should only [need to] save the
>> base-name to disk, so there's no need to change the disk representation
>> of your data.

> The "base name" and the buffer name are two different pieces of info;
> even if no uniquify is present, it would be perhaps interesting to
> know (and save) whether "my-buffer<2>" is called so by user fiat or
> via rename-buffer's UNIQUE argument; but that is not needed right now.

> Anyway. Does this version lessen your yucky factor somewhat?

Yes, very much so.

> +(defun uniquify-buffer-base-name ()
> +  "Return the base name of the current buffer."
> +  (and (bound-and-true-p uniquify-managed)
> +       (uniquify-item-base (car uniquify-managed))))

There's a (defvar uniquify-managed nil ...) earlier in that file, so you
shouldn't need the `bound-and-true-p', just use

   (defun uniquify-buffer-base-name ()
     "Return the base name of the current buffer.
   Return nil if the buffer is not managed by uniquify."
     (and uniquify-managed
          (uniquify-item-base (car uniquify-managed))))


-- Stefan

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Feb 12, 2008 4:39 AM, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> Yes, very much so.

Glad to hear.

> > +(defun uniquify-buffer-base-name ()
> > +  "Return the base name of the current buffer."
> > +  (and (bound-and-true-p uniquify-managed)
> > +       (uniquify-item-base (car uniquify-managed))))
>
> There's a (defvar uniquify-managed nil ...) earlier in that file, so you
> shouldn't need the `bound-and-true-p', just use

Hmm.  IIUC, you're saying than if `uniquify-buffer-base-name' is run,
then uniquify.el has been loaded and so `uniquify-managed' will be
bound, even if it has no buffer-local value for the buffer being
tested... You're right. I was thinking of cases when uniquify.el were
not loaded, but they don't apply here.

>    (defun uniquify-buffer-base-name ()
>      "Return the base name of the current buffer.
>    Return nil if the buffer is not managed by uniquify."

OK.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Martin Fischer]

  >>>>> Martin Fischer writes:
  > Date: Wed, 06 Feb 2008 17:12:31 +0100

8<--------------------------------------------------------------------

  >   ...\emacs-22.1.90\bin\runemacs.exe -Q
  > 
  >   Customize the variable uniquify-buffer-name-style to "reverse".
  > 
  >   List the directories c:\otto\ludwig and c:\karl\ludwig with dired.
  > 
  >   M-x toggle-debug-on-error RET
  > 
  >   M-x desktop-save RET
  > 
  >     Debugger entered--Lisp error: (error "No buffer named ludwig")
  >       desktop-save-buffer-p(nil "ludwig" dired-mode nil 185 (nil
  >       nil) t ("c:/otto/ludwig/") nil)
  >       apply(desktop-save-buffer-p (nil "ludwig" dired-mode nil 185
  >       (nil nil) t ("c:/otto/ludwig/") nil))
  >       desktop-save("~/")
  >       call-interactively(desktop-save)
  >       execute-extended-command(nil)
  >       call-interactively(execute-extended-command)
  > 
  >   M-x list-buffers RET
  > 
  >     .%* *Backtrace*            392  Debugger
  >      %  *Completions*          167  Completion List
  >      %  ludwig\karl\           184  Dired by name	  c:/karl/ludwig/
  >      %  ludwig\otto\           184  Dired by name	  c:/otto/ludwig/
  >         *scratch*                0  Lisp Interaction
  >       * *Messages*             957  Fundamental

8<--------------------------------------------------------------------

Hi,

thanks to Stefan and Juanma for solving this problem in the new
pretest version (emacs-22.1.91).

Nevertheless the solution opened up a new one:

  If I look at the root directory of a windows partition in dired mode
  (uniquify-managed), desktop-save creates an entry with an empty
  buffer name (uniquify-buffer-base-name seems to return that) in the
  .emacs.desktop file:

    (desktop-create-buffer 206
      nil
  --> ""
      'dired-mode
      nil
      118
      '(nil nil)
      t
      '("c:/")
      nil)

  This in turn prevents desktop-read to continue processing the rest of
  the .emacs.desktop file after issuing the message

    "setq: Empty string is invalid as a buffer name".

  .emacs:

    (custom-set-variables
      ;; custom-set-variables was added by Custom.
      ;; If you edit it by hand, you could mess it up, so be careful.
      ;; Your init file should contain only one such instance.
      ;; If there is more than one, they won't work right.
     '(desktop-save-mode t)
     '(uniquify-buffer-name-style (quote reverse) nil (uniquify)))
    (custom-set-faces
      ;; custom-set-faces was added by Custom.
      ;; If you edit it by hand, you could mess it up, so be careful.
      ;; Your init file should contain only one such instance.
      ;; If there is more than one, they won't work right.
     )

Sorry for being a nuisance.

Martin
-- 
parozusa at web dot de

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Wed, Feb 27, 2008 at 1:43 PM, Martin Fischer <parozusa@web.de> wrote:

>   If I look at the root directory of a windows partition in dired mode
>   (uniquify-managed), desktop-save creates an entry with an empty
>   buffer name (uniquify-buffer-base-name seems to return that) in the
>   .emacs.desktop file:
>
>     (desktop-create-buffer 206
>       nil
>   --> ""

Aha. I'll look into it ASAP.

>  Sorry for being a nuisance.

Thanks for being thorough.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Wed, Feb 27, 2008 at 1:48 PM, Juanma Barranquero <lekktu@gmail.com> wrote:

>  Aha. I'll look into it ASAP.

Please, try updating uniquify.el from the EMACS_22_BRANCH, or applying
the following patch if you're using a pretest tarball.

             Juanma


Index: lisp/uniquify.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/uniquify.el,v
retrieving revision 1.62.2.8
retrieving revision 1.62.2.9
diff -u -2 -r1.62.2.8 -r1.62.2.9
--- lisp/uniquify.el	12 Feb 2008 14:32:56 -0000	1.62.2.8
+++ lisp/uniquify.el	27 Feb 2008 16:20:33 -0000	1.62.2.9
@@ -195,5 +195,6 @@
 Return nil if the buffer is not managed by uniquify."
   (and uniquify-managed
-       (uniquify-item-base (car uniquify-managed))))
+       (let ((base (uniquify-item-base (car uniquify-managed))))
+	 (if (string= base "") nil base))))

 ;;; Main entry point.

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Stefan Monnier]

> @@ -195,5 +195,6 @@
>  Return nil if the buffer is not managed by uniquify."
>    (and uniquify-managed
> -       (uniquify-item-base (car uniquify-managed))))
> +       (let ((base (uniquify-item-base (car uniquify-managed))))
> +	 (if (string= base "") nil base))))

Hmm... but then that doesn't seem to agree with the docstring:
the buffer *is* managed by uniquify, but you return nil.

I think the fix should be in desktop.


        Stefan

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

>  Hmm... but then that doesn't seem to agree with the docstring:
>  the buffer *is* managed by uniquify, but you return nil.

I changed the docstring in the very next commit.

>  I think the fix should be in desktop.

I don't think so. The function is not called
uniquify-buffer-managed-by-uniquify-p; it's an function whose whole
point is to provide a nicer interface for desktop, and, to desktop, a
basename of "" is like having no basename.

If we ever find that the function is useful for other packages we can
revisit the issue.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Stefan Monnier]

>> I think the fix should be in desktop.

> I don't think so. The function is not called
> uniquify-buffer-managed-by-uniquify-p; it's an function whose whole
> point is to provide a nicer interface for desktop,

No, the whole point of the function is to provide a meaningful
functionality independently from desktop.el.  If it's specific to
desktop.el, then it defeats the whole purpose: you may as well go back
to looking at `uniquify-managed' directly from desktop.el.

> and, to desktop, a basename of "" is like having no basename.

Exactly: *to desktop* this is so, but not in general, so the test
belongs in desktop.el.


        Stefan

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Wed, Feb 27, 2008 at 11:37 PM, Stefan Monnier
<monnier@iro.umontreal.ca> wrote:

>  No, the whole point of the function is to provide a meaningful
>  functionality independently from desktop.el.

As there are no other users, we don't know if it is more useful and
general to return nil for empty basenames or to return "".

>  Exactly: *to desktop* this is so, but not in general, so the test
>  belongs in desktop.el.

There are no general uses of `uniquify-buffer-base-name'. There is no
general definition of a "base name", it is an internal thing of
uniquify; so it is up to us to decide whether an empty basename is a
base name at all or not. For the moment being, the only user (desktop)
is cleaner if an empty basename and no basename at all are equivalent.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Stefan Monnier]

>> No, the whole point of the function is to provide a meaningful
>> functionality independently from desktop.el.
> As there are no other users, we don't know if it is more useful and
> general to return nil for empty basenames or to return "".

We know that one is meaningful: it indicates that the buffer name *is*
under uniquify's control.

>> Exactly: *to desktop* this is so, but not in general, so the test
>> belongs in desktop.el.

> There are no general uses of `uniquify-buffer-base-name'. There is no
> general definition of a "base name", it is an internal thing of
> uniquify;

It's not linked to its implementation but to its featureset, so it's
"specific" to uniquify, but not "internal".  Which is why it's a good
interface which stands on its own, whether or not desktop.el is its
only user.

Please undo your change to uniquify and move the test to desktop.el.


        Stefan

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Thu, Feb 28, 2008 at 2:54 AM, Stefan Monnier
<monnier@iro.umontreal.ca> wrote:

>  Please undo your change to uniquify and move the test to desktop.el.

I am not convinced. You want to change it, change it.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Stefan Monnier]

>> Please undo your change to uniquify and move the test to desktop.el.

> I am not convinced. You want to change it, change it.

I've undone your change to uniquify.


        Stefan

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Martin Fischer]

  >>>>> Martin Fischer writes:
  > Date: Wed, 27 Feb 2008 13:43:22 +0100

8<--------------------------------------------------------------------

  > Nevertheless the solution opened up a new one:
  > 
  >   If I look at the root directory of a windows partition in dired mode
  >   (uniquify-managed), desktop-save creates an entry with an empty
  >   buffer name (uniquify-buffer-base-name seems to return that) in the
  >   .emacs.desktop file:

8<--------------------------------------------------------------------

Hi,

I wasn't thorough enough ;-), it isn't a new problem.

It already exists in

  GNU Emacs 22.1.50.1 (i386-mingw-nt5.1.2600) of 2007-08-18 on TPAD

where desktop doesn't depend on uniquify.

Cheers

Martin
-- 
parozusa at web dot de

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Thu, Feb 28, 2008 at 2:27 PM, Martin Fischer <parozusa@web.de> wrote:

>  I wasn't thorough enough ;-), it isn't a new problem.
>
>  It already exists in
>
>   GNU Emacs 22.1.50.1 (i386-mingw-nt5.1.2600) of 2007-08-18 on TPAD
>
>  where desktop doesn't depend on uniquify.

Could you elaborate, please?

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Martin Fischer]

  >>>>> Juanma Barranquero writes:
  > Date: Thu, 28 Feb 2008 17:29:15 +0100
  > 
  > On Thu, Feb 28, 2008 at 2:27 PM, Martin Fischer wrote:
  > 
  >> I wasn't thorough enough ;-), it isn't a new problem.
  >> 
  >> It already exists in
  >> 
  >> GNU Emacs 22.1.50.1 (i386-mingw-nt5.1.2600) of 2007-08-18 on TPAD
  >> 
  >> where desktop doesn't depend on uniquify.
  > 
  > Could you elaborate, please?
  > 
  >              Juanma

Hi,

being thorough again:

  - Customize uniquify-buffer-name-style.

      (custom-set-variables
       '(uniquify-buffer-name-style (quote reverse) nil (uniquify)))

  - Create buffers for some file and a root directory and save the
    configuration.

      M-x find-file RET c:/AUTOEXEC.BAT RET
      M-x dired RET c:/ RET
      M-x desktop-save RET c:/ RET

  - look at .emacs.desktop:

      (desktop-create-buffer 206
        nil
    --> ""
        'dired-mode
        nil
        118
        '(nil nil)
        t
        '("c:/")
        nil)

      (desktop-create-buffer 206
        "c:/AUTOEXEC.BAT"
        "AUTOEXEC.BAT"
        'fundamental-mode
        nil
        1
        '(nil nil)
        nil
        nil
        nil)

  - if you remove both buffers and try to restore them by reading
    .emacs.desktop with desktop-read, you get

      "setq: Empty string is invalid as a buffer name"

    in the *Message* buffer.

    Nonetheless the dired buffer is created correctly, but not the other
    one, which means processing of .emacs.desktop stopped.

  - Here is the backtrace after toggle-debug-on-error:

    Debugger entered--Lisp error: (error "Empty string is invalid as a buffer name")
      ad-Orig-rename-buffer("" t)
      (setq ad-return-value (ad-Orig-rename-buffer newname unique))
      (let (ad-return-value) (setq ad-return-value (ad-Orig-rename-buffer newname unique)) (uniquify-maybe-rerationalize-w/o-cb) (if (null unique) (setq uniquify-managed nil) (when uniquify-buffer-name-style ... ...)) ad-return-value)
      rename-buffer("" t)
      desktop-create-buffer(206 nil "" dired-mode nil 118 (nil nil) t ("c:/") nil)
      eval-buffer(#<buffer  *load*> nil "c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" nil t)  ; Reading at buffer position 690
      load-with-code-conversion("c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" "c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" t t)
      load("c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" t t t)
      desktop-read()
      call-interactively(desktop-read)
      execute-extended-command(nil)
      call-interactively(execute-extended-command)

Cheers

Martin
-- 
parozusa at web dot de

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Fri, Feb 29, 2008 at 11:07 AM, Martin Fischer <parozusa@web.de> wrote:

>  being thorough again:

I cannot reproduce it with either EMACS_22_BASE or the trunk. Both
work as expected.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Martin Fischer]

[-- Attachment #1: Type: text/plain, Size: 1800 bytes --]


  >>>>> Juanma Barranquero writes:
  > Date: Fri, 29 Feb 2008 11:49:50 +0100
  > 
  > On Fri, Feb 29, 2008 at 11:07 AM, Martin Fischer <parozusa@web.de> wrote:
  > 
  >> being thorough again:
  > 
  > I cannot reproduce it with either EMACS_22_BASE or the trunk. Both
  > work as expected.
  > 
  >              Juanma

Hi,

it's

  GNU Emacs 22.1.50.1 (i386-mingw-nt5.1.2600) of 2007-08-18 on TPAD.

I got it from

  http://ntemacs.sourceforge.net/

I call it like

  $ d:/aemf/ntemacs22/bin/runemacs.exe -Q

from 

  MKS Bash, version 8.7.0 build 1859 Apr 16 2004 10:24:01 (x86-MKS-Windows_NT)

on WindowsXP.

I evaluate the attached .emacs and M-x toggle-debug-on-error RET.

M-x desktop-read RET on the attached .emacs.desktop gives me the
following backtrace:

Debugger entered--Lisp error: (error "Empty string is invalid as a buffer name")
  ad-Orig-rename-buffer("" t)
  (setq ad-return-value (ad-Orig-rename-buffer newname unique))
  (let (ad-return-value) (setq ad-return-value (ad-Orig-rename-buffer newname unique)) (uniquify-maybe-rerationalize-w/o-cb) (if (null unique) (setq uniquify-managed nil) (when uniquify-buffer-name-style ... ...)) ad-return-value)
  rename-buffer("" t)
  desktop-create-buffer(206 nil "" dired-mode nil 118 (nil nil) t ("c:/") nil)
  eval-buffer(#<buffer  *load*> nil "c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" nil t)  ; Reading at buffer position 690
  load-with-code-conversion("c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" "c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" t t)
  load("c:/Dokumente und Einstellungen/aemf/empty/.emacs.desktop" t t t)
  desktop-read()
  call-interactively(desktop-read)
  execute-extended-command(nil)
  call-interactively(execute-extended-command)

Martin
-- 
parozusa at web dot de

[-- Attachment #2: .emacs --]
[-- Type: application/octet-stream, Size: 568 bytes --]

(custom-set-variables
  ;; custom-set-variables was added by Custom.
  ;; If you edit it by hand, you could mess it up, so be careful.
  ;; Your init file should contain only one such instance.
  ;; If there is more than one, they won't work right.
 '(uniquify-buffer-name-style (quote reverse) nil (uniquify)))
(custom-set-faces
  ;; custom-set-faces was added by Custom.
  ;; If you edit it by hand, you could mess it up, so be careful.
  ;; Your init file should contain only one such instance.
  ;; If there is more than one, they won't work right.
 )

[-- Attachment #3: .emacs.desktop --]
[-- Type: application/octet-stream, Size: 719 bytes --]

;; -*- mode: emacs-lisp; coding: emacs-mule; -*-
;; --------------------------------------------------------------------------
;; Desktop File for Emacs
;; --------------------------------------------------------------------------
;; Created Fri Feb 29 13:39:15 2008
;; Desktop file format version 206
;; Emacs version 22.1.50.1

;; Global section:
(setq desktop-missing-file-warning nil)
(setq tags-file-name nil)
(setq tags-table-list nil)
(setq search-ring nil)
(setq regexp-search-ring nil)
(setq register-alist nil)

;; Buffer section -- buffers listed in same order as in buffer list:
(desktop-create-buffer 206
  nil
  ""
  'dired-mode
  nil
  118
  '(nil nil)
  t
  '("c:/")
  nil)

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Fri, Feb 29, 2008 at 2:17 PM, Martin Fischer <parozusa@web.de> wrote:

>   GNU Emacs 22.1.50.1 (i386-mingw-nt5.1.2600) of 2007-08-18 on TPAD.
>
>  I got it from
>
>   http://ntemacs.sourceforge.net/

Well, yes. This is an unofficial build of Emacs, apparently checked
out from the CVS before I fixed the bug in your original report. It's
no wonder that it fails...

If you want a more up-to-date binary, perhaps you can write to the
maintainer of that port, or try at
http://ourcomments.org/Emacs/EmacsW32.html

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Jason Rumney]

Juanma Barranquero wrote:
> If you want a more up-to-date binary, perhaps you can write to the
> maintainer of that port, or try at
> http://ourcomments.org/Emacs/EmacsW32.html
>   

The one there is even older. I think the maintainers of unofficial CVS 
builds lost interest when 22.1 was released. Maybe in a few years they 
will pick back up again if we still have not released 23.1.

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Fri, Feb 29, 2008 at 3:48 PM, Jason Rumney <jasonr@gnu.org> wrote:

>  The one there is even older.

My mistake. I though Lennart was building from CVS regularly.

> I think the maintainers of unofficial CVS
>  builds lost interest when 22.1 was released. Maybe in a few years they
>  will pick back up again if we still have not released 23.1.

Perhaps we should provide non-official weekly builds or somesuch.

             Juanma

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Lennart Borgman (gmail)]

Juanma Barranquero wrote:
> On Fri, Feb 29, 2008 at 3:48 PM, Jason Rumney <jasonr@gnu.org> wrote:
> 
>>  The one there is even older.
> 
> My mistake. I though Lennart was building from CVS regularly.

I did, but I gave up because I did not have time to take care of some 
changes that broke my patches. Then I have been too busy making nXhtml 
and some other things to look at this again.

But don't worry, I will try again.

BTW, the emacsclient patches are part of those patches. And of course, 
the patches that allows me to use the Windows keys as META.

>> I think the maintainers of unofficial CVS
>>  builds lost interest when 22.1 was released. Maybe in a few years they
>>  will pick back up again if we still have not released 23.1.
> 
> Perhaps we should provide non-official weekly builds or somesuch.
> 
>              Juanma
> 
> 
> 

Re: 22.1.90; desktop-save throws (error "No buffer named <uniquified buffername>")

[Juanma Barranquero]

On Fri, Feb 29, 2008 at 4:16 PM, Lennart Borgman (gmail)
<lennart.borgman@gmail.com> wrote:

>  But don't worry, I will try again.

Glad to hear.

>  BTW, the emacsclient patches are part of those patches. And of course,
>  the patches that allows me to use the Windows keys as META.

Wouldn't even dream otherwise.

             Juanma