bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

[Reiner Steib]

Please describe exactly what actions triggered the bug
and the precise symptoms of the bug.  If you can, give
a recipe starting from `emacs -Q':

(a) emacs -Q -eval \
     "(defalias 'tramp-completion-file-name-handler 'file-name-non-special)"

     I added this alias because I couldn't prevent tramp kicking in
     (which adds quite a long delay the first time tramp is loaded) by
     simply customizing `tramp-mode' to nil, as I expected to from the
     doc string. Then I found that `file-name-handler-alist' has entries
     for `tramp-completion-file-name-handler' and
     `tramp-file-name-handler', and concluded that aliasing these to the
     default `file-name-non-special' might do the trick.

(b) M-x cvs-update RET

(c) I get a prompt like this:

     "Buffer  *temp* modified; kill anyway? "

(d) With M-x toggle-debug-on-quit RET, I get the following:

    Debugger entered--Lisp error: (quit)
      yes-or-no-p("Buffer  *temp*<4> modified; kill anyway? ")
      kill-buffer(#<buffer  *temp*<4>>)
      byte-code("[...]" [temp-buffer buffer-name kill-buffer] 2)
      byte-code([...] [temp-buffer cvs-cvsrc-file cmd --cl-dolist-temp-- sym
        val generate-new-buffer " *temp*" ((byte-code "..." [temp-buffer
        buffer-name kill-buffer] 2)) insert-file-contents ("cvs" "checkout"
        "status" "log" "diff" "tag" "add" "commit" "remove" "update") nil
        re-search-forward "^" "\\(\\s-+\\(.*\\)\\)?$" t intern "cvs-"
        "-flags" split-string-and-unquote match-string 2 "" 0 error
        "cvs-flags-defaults accessing a non-cvs-flags" 1 cvs-cvs-flags "-f"
        cvs-partition #[(x) "..." [x ("-q" "-Q" "-f")] 2] cvs-flags-query
        noquery value index cl-struct-cvs-flags-tags] 9)
      cvs-reread-cvsrc()
      execute-extended-command(nil)
      call-interactively(execute-extended-command nil nil)


In GNU Emacs 24.0.50.1 (i386-mingw-nt6.0.6002)
  of 2011-01-10 on 3249CTO
Windowing system distributor `Microsoft Corp.', version 6.0.6002
configured using `configure --with-gcc (4.4) --no-opt --cflags
-Ic:/imagesupport/include'

(I have the same behavior with 23.2)

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
   value of $XMODIFIERS: nil
   locale-coding-system: cp1252
   default enable-multibyte-characters: t

Major mode: Debugger

Minor modes in effect:
   tooltip-mode: t
   mouse-wheel-mode: t
   tool-bar-mode: t
   menu-bar-mode: t
   file-name-shadow-mode: t
   global-font-lock-mode: t
   font-lock-mode: t
   blink-cursor-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-x c v s - u p <tab> <return> C-g M-x t o g g l e
- d e b u g - o n - q u i t <return> M-x <up> <up>
<return> C-g M-x r e p o r t - e m a c s - b u g <
return>

Recent messages:
For information about GNU Emacs and the GNU system, type C-h C-a.
Quit
Debug on Quit enabled globally
Entering debugger...

Load-path shadows:
None found.

Features:
(shadow sort gnus-util mail-extr message rfc822 mml mml-sec mm-decode
mm-bodies mm-encode mail-parse rfc2231 rfc2047 rfc2045 ietf-drums
mm-util mail-prsvr mailabbrev mail-utils gmm-utils mailheader emacsbug
debug cus-start cus-load pcvs-parse pcvs-info pcvs-defs easymenu
easy-mmode pcvs-util ewoc tooltip ediff-hook vc-hooks lisp-float-type
mwheel dos-w32 disp-table ls-lisp w32-win w32-vars tool-bar dnd fontset
image fringe lisp-mode register page menu-bar rfn-eshadow timer select
scroll-bar mouse jit-lock font-lock syntax facemenu font-core frame cham
georgian utf-8-lang misc-lang vietnamese tibetan thai tai-viet lao
korean japanese hebrew greek romanian slovak czech european ethiopic
indian cyrillic chinese case-table epa-hook jka-cmpr-hook help simple
abbrev button minibuffer faces cus-face files text-properties overlay
md5 base64 format env code-pages mule custom widget
hashtable-print-readable backquote make-network-process multi-tty emacs)

bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

[Stefan Monnier]

> Please describe exactly what actions triggered the bug
> and the precise symptoms of the bug.  If you can, give
> a recipe starting from `emacs -Q':

> (a) emacs -Q -eval \
>     "(defalias 'tramp-completion-file-name-handler 'file-name-non-special)"

>     I added this alias because I couldn't prevent tramp kicking in
>     (which adds quite a long delay the first time tramp is loaded) by
>     simply customizing `tramp-mode' to nil, as I expected to from the
>     doc string. Then I found that `file-name-handler-alist' has entries
>     for `tramp-completion-file-name-handler' and
>     `tramp-file-name-handler', and concluded that aliasing these to the
>     default `file-name-non-special' might do the trick.

> (b) M-x cvs-update RET

> (c) I get a prompt like this:

>     "Buffer  *temp* modified; kill anyway? "

Hmm... can't reproduce this here.  I get a prompt "CVS Update (directory):"
instead, and no matter what directory I select I don't get this bug.


        Stefan

bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

[Reiner Steib]

On Mon, Jan 17 2011, Stefan Monnier wrote:

>> (a) emacs -Q -eval \
>>     "(defalias 'tramp-completion-file-name-handler 'file-name-non-special)"
>
>>     I added this alias because I couldn't prevent tramp kicking in
>>     (which adds quite a long delay the first time tramp is loaded) by
>>     simply customizing `tramp-mode' to nil, as I expected to from the
>>     doc string. Then I found that `file-name-handler-alist' has entries
>>     for `tramp-completion-file-name-handler' and
>>     `tramp-file-name-handler', and concluded that aliasing these to the
>>     default `file-name-non-special' might do the trick.
>
>> (b) M-x cvs-update RET
>
>> (c) I get a prompt like this:
>
>>     "Buffer  *temp* modified; kill anyway? "
>
> Hmm... can't reproduce this here.  I get a prompt "CVS Update (directory):"
> instead, and no matter what directory I select I don't get this bug.

The bug happens before I can select the directory, so the specific
directory doesn't matter.

I neither can reproduce it on GNU/Linux, but on Windows, it is 100%
reproducible.  Any ideas how I can obtain more useful information from
the debug buffer?  Which *.el file should I load?

One more data point: The problem also happens e.g. when starting
Gnus.  My guess is that it happen when Emacs reads a file (.cvsrc,
...) into a temp buffer.

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

[Reiner Steib]

Reiner Steib writes:

> The bug happens before I can select the directory, so the specific
> directory doesn't matter.
> 
> I neither can reproduce it on GNU/Linux, but on Windows, it is 100%
> reproducible.  Any ideas how I can obtain more useful information from
> the debug buffer?  Which *.el file should I load?
> 
> One more data point: The problem also happens e.g. when starting
> Gnus.  My guess is that it happen when Emacs reads a file (.cvsrc,
> ...) into a temp buffer.

After moving ~/.cvsrc away, the bug doesn't happen. 
With a 0 byte ~/.cvsrc, the bug doesn't happen. 

With this simple .cvsrc, the bug triggers again:

$ cat --show-all ~/.cvsrc
#

(i.e. only one character '#', no CR, no newline).

When I answer "no" at the prompt "Buffer  *temp* modified; kill anyway?",
visit the buffer " *temp*", (point-max) is 2, and the mode line shows
iso-latin-1-dos as the coding system.

M-x describe-coding-system shows RET:

| Coding system for saving this buffer:
|   Not set locally, use the default.
| Default coding system (for new files):
|   1 -- iso-latin-1-dos (alias: iso-8859-1-dos latin-1-dos)

Bye, Reiner.

bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

[Stefan Monnier]

> After moving ~/.cvsrc away, the bug doesn't happen. 
> With a 0 byte ~/.cvsrc, the bug doesn't happen. 

> With this simple .cvsrc, the bug triggers again:

> $ cat --show-all ~/.cvsrc
> #

> (i.e. only one character '#', no CR, no newline).

So I guess the problem can be reproduced with

  emacs -Q --eval \
      "(defalias 'tramp-completion-file-name-handler 'file-name-non-special)"

and then M-x load-library RET pcvs RET.

I think I see the source of the problem.  In file-name-non-special we
do:

	(file-arg-indices
	 (cdr (or (assq operation
			;; The first six are special because they
			;; return a file name.  We want to include the /:
			;; in the return value.
			;; So just avoid stripping it in the first place.
			'((expand-file-name . nil)
                          [...]
			  ;; `quote' means add "/:" to buffer-file-name.
			  (insert-file-contents quote 0)
                          [...]
			  (add-name-to-file 0 1)))
          [...]
	  ((eq method 'quote)
	   (unwind-protect
	       (apply operation arguments)
	     (setq buffer-file-name (concat "/:" buffer-file-name))))

and indeed pcvs.el's cvs-read-cvsrc calls insert-file-contents, but not
in a way that should set buffer-file-name (i.e. it doesn't set the
`visit' argument).
I still don't understand why this shows up under w32 and not under
GNU/Linux, but I'm pretty sure that's the culprit.  Can you check to see
if the quick-fix below solves the problem?

             
        Stefan


=== modified file 'lisp/files.el'
--- lisp/files.el	2011-01-18 01:38:22 +0000
+++ lisp/files.el	2011-01-20 15:21:55 +0000
@@ -6165,7 +6165,8 @@
 	  ((eq method 'quote)
 	   (unwind-protect
 	       (apply operation arguments)
-	     (setq buffer-file-name (concat "/:" buffer-file-name))))
+             (if buffer-file-name
+                 (setq buffer-file-name (concat "/:" buffer-file-name)))))
 	  ((eq method 'unquote-then-quote)
 	   (let (res)
 	     (setq buffer-file-name (substring buffer-file-name 2))

bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

[Reiner Steib]

On Thu, Jan 20 2011, Stefan Monnier wrote:

> So I guess the problem can be reproduced with
>
>   emacs -Q --eval \
>       "(defalias 'tramp-completion-file-name-handler 'file-name-non-special)"
>
> and then M-x load-library RET pcvs RET.

Yes, that is another way I can reproduce it.

> I still don't understand why this shows up under w32 and not under
> GNU/Linux, but I'm pretty sure that's the culprit.  

Maybe a line-endings issue?

> Can you check to see if the quick-fix below solves the problem?

It solves the problem.  Thank you!

> -	     (setq buffer-file-name (concat "/:" buffer-file-name))))
> +             (if buffer-file-name
> +                 (setq buffer-file-name (concat "/:" buffer-file-name)))))

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?

[Stefan Monnier]

>> Can you check to see if the quick-fix below solves the problem?
> It solves the problem.  Thank you!

Thanks for confirming.  I've installed the patch below in the emacs-23
branch, which should fix it right.


        Stefan


=== modified file 'lisp/files.el'
--- lisp/files.el	2011-01-20 02:55:36 +0000
+++ lisp/files.el	2011-01-24 20:22:34 +0000
@@ -6118,8 +6118,7 @@
 			  (substitute-in-file-name identity)
 			  ;; `add' means add "/:" to the result.
 			  (file-truename add 0)
-			  ;; `quote' means add "/:" to buffer-file-name.
-			  (insert-file-contents quote 0)
+			  (insert-file-contents insert-file-contents 0)
 			  ;; `unquote-then-quote' means set buffer-file-name
 			  ;; temporarily to unquoted filename.
 			  (verify-visited-file-modtime unquote-then-quote)
@@ -6150,20 +6149,18 @@
 			   "/"
 			 (substring (car pair) 2)))))
 	(setq file-arg-indices (cdr file-arg-indices))))
-    (cond ((eq method 'identity)
-	   (car arguments))
-	  ((eq method 'add)
-	   (concat "/:" (apply operation arguments)))
-	  ((eq method 'quote)
-	   (unwind-protect
+    (case method
+      (identity (car arguments))
+      (add (concat "/:" (apply operation arguments)))
+      (insert-file-contents
+       (let ((visit (nth 1 arguments)))
+         (prog1
 	       (apply operation arguments)
-	     (setq buffer-file-name (concat "/:" buffer-file-name))))
-	  ((eq method 'unquote-then-quote)
-	   (let (res)
-	     (setq buffer-file-name (substring buffer-file-name 2))
-	     (setq res (apply operation arguments))
-	     (setq buffer-file-name (concat "/:" buffer-file-name))
-	     res))
+           (when (and visit buffer-file-name)
+             (setq buffer-file-name (concat "/:" buffer-file-name))))))
+      (unquote-then-quote
+       (let ((buffer-file-name (substring buffer-file-name 2)))
+         (apply operation arguments)))
 	  (t
 	   (apply operation arguments)))))
 \f

bug#7854: emacs-23 (regression) - fix breaks `dired' (was: bug#7854: 24.0.50; Buffer *temp* modified; kill anyway?)

[Reiner Steib]

Stefan Monnier <monnier <at> iro.umontreal.ca> writes:

> 
> >> Can you check to see if the quick-fix below solves the problem?
> > It solves the problem.  Thank you!
> 
> Thanks for confirming.  I've installed the patch below in the emacs-23
> branch, which should fix it right.

This is the relevant commit:

+2011-01-24 Stefan Monnier <monnier@iro.umontreal.ca>
+
+ * files.el (file-name-non-special): Only change buffer-file-name after
+ insert-file-contents if it's `visit'ing the file (bug#7854).

It fixed the original problem, but there's a regression: It breaks `dired' for
me (not sure if I should have opened a new bug?):

$ emacs -Q -l files-emacs23-git.el --eval '(dired "u:/tmp/")'
(files-emacs23-git.el =
http://git.savannah.gnu.org/cgit/emacs.git/plain/lisp/files.el?h=emacs-23&id=8588a5a723f4e5cfa115341a242126a9efc71808)

I get:

| if: Reading directory: "ls -al -- u:/tmp/" exited with status 1

$ emacs -Q -l files-emacs23-git.el -f toggle-debug-on-error \
    --eval '(dired "u:/tmp/")'

| dired-readin()
| dired-internal-noselect("u:/tmp/" nil)
| dired-noselect("u:/tmp/" nil)
| dired("u:/tmp/")
| eval((dired "u:/tmp/"))
| command-line-1(("-l" "files-emacs23-git.el" "-f"
|   "toggle-debug-on-error" "--eval" "(dired \"u:/tmp/\")"))
| command-line()
| normal-top-level()

Edebuging `dired-internal-noselect' ...

| edebug-enter(dired-internal-noselect ("u:/tmp/" nil nil) (lambda nil
|   (edebug-after (edebug-before 0) 121 (let*
|   ... ... ... ... ... ...))))
| dired-internal-noselect("u:/tmp/" nil)
| dired-noselect("u:/tmp/" nil)
| dired("u:/tmp/" nil)
| call-interactively(dired t nil)
| execute-extended-command(nil)
| call-interactively(execute-extended-command nil nil)
| recursive-edit()
| byte-code("..." [unread-command-char debugger-args x debugger-buffer
|   noninteractive debugger-batch-max-lines -1 debug backtrace-debug 4 t
|   backtrace-frame lambda 5 pop-to-buffer debugger-mode
|   debugger-setup-buffer count-lines 2 "...\n" message "%s"
|   buffer-string kill-emacs "" nil recursive-edit middlestart
|   buffer-read-only standard-output] 4)
| debug(error (error "Reading directory: \"ls -al -- u:/tmp/\" exited
|   with status 1"))
| signal(error ("Reading directory: \"ls -al -- u:/tmp/\" exited with
|   status 1"))
| error("Reading directory: \"%s %s -- %s\" exited with status %s"
|   "ls" "-al" "u:/tmp/" 1)
| (if (and (file-directory-p file) (memq system-type ...)) (error
|   "Reading directory: \"%s %s -- %s\" exited with status %s"
|   insert-directory-program (if ... ... switches) file result)
|   (access-file file "Reading directory") (error "Listing directory
|   failed but `access-file' worked"))
| (if (eq 0 result) nil (delete-region beg (point)) (if (and ... ...)
|   (error "Reading directory: \"%s %s -- %s\" exited with status %s"
|   insert-directory-program ... file result) (access-file file "Reading
|   directory") (error "Listing directory failed but `access-file'
|   worked")))
| (unless (eq 0 result) (delete-region beg (point)) (if (and ... ...)
|   (error "Reading directory: \"%s %s -- %s\" exited with status %s"
|   insert-directory-program ... file result) (access-file file "Reading
|   directory") (error "Listing directory failed but `access-file'
|   worked")))
| (let (result (beg ...)) (let* (... ...) (setq result ...)) (when (if
|   ... ... ...) (save-excursion ... ... ...)) (when (and ... ...) (let
|   ... ...)) (when (and ... ...) (setq result 0)) (unless (eq 0 result)
|   (delete-region beg ...) (if ... ... ... ...)) (when (if ... ... ...)
|   (forward-line -2) (when ... ... ...) (if ... ... ...) (if ... ...))
|   (let (... coding-no-eol val pos) (when ... ... ...)) (if
|   full-directory-p (save-excursion ... ...)))
| (if handler (funcall handler (quote insert-directory) file switches
|   wildcard full-directory-p) (let (result ...) (let* ... ...) (when
|   ... ...) (when ... ...) (when ... ...) (unless ... ... ...) (when
|   ... ... ... ... ...) (let ... ...) (if full-directory-p ...)))
| (let ((handler ...)) (if handler (funcall handler ... file switches
|   wildcard full-directory-p) (let
|   ... ... ... ... ... ... ... ... ...)))
| insert-directory("u:/tmp/" "-al" nil t)
| dired-insert-directory("u:/tmp/" "-al" nil nil t)
| dired-readin-insert()
| dired-readin()
| dired-internal-noselect("u:/tmp/" nil)
| dired-noselect("u:/tmp/" nil)
| dired("u:/tmp/")
| eval((dired "u:/tmp/"))
| command-line-1(("-l" "files-emacs23-git.el" "-f"
|   "toggle-debug-on-error" "--eval" "(dired \"u:/tmp/\")"))
| command-line()
| normal-top-level()

| $ u:/bin/ls --version
| ls (fileutils) 4.4.219 2010/05
| Written by Richard Stallman and David MacKenzie
| Microsoft Windows extensions by Alan Klietz.
| 
| Copyright (C) 2001 Free Software Foundation, Inc.
| 
| Microsoft Windows extensions copyright (C) 2010, Algin Technology LLC
| Distributed under GNU General Public License version 2.
| 
| This is free software; see the source for copying conditions.  There is NO
| warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Executing (shell-command-to-string "ls -al u:/tmp/") in IELM gives
resonable a result.

bug#7854: emacs-23 (regression) - fix breaks `dired'

[Chong Yidong]

Could someone with access to Windows please test the regression reported
at Bug#7854?  I could not reproduce it, replacing "u:/tmp" with "/tmp/".

> $ emacs -Q -l files-emacs23-git.el --eval '(dired "u:/tmp/")'
> (files-emacs23-git.el =
> http://git.savannah.gnu.org/cgit/emacs.git/plain/lisp/files.el?h=emacs-23&id=8588a5a723f4e5cfa115341a242126a9efc71808)
>
> I get:
>
> | if: Reading directory: "ls -al -- u:/tmp/" exited with status 1

If someone could come up with a smaller testcase, that would be good
too.