23.0.60; M-x compile gives args out of range 0, 0

23.0.60; M-x compile gives args out of range 0, 0

[Jan Djärv]

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:

When doing M-x compile I get:
font-lock-extend-region-multiline: Args out of range: 0, 0

It seems to be random, sometimes I can compile in other directories, sometimes
not.  But args out of range comes about 3 times out of 4.  It seems to always
come in the emacs/src directory.

Debug-on-error t gives:

Debugger entered--Lisp error: (args-out-of-range 0 0)
   get-text-property(0 font-lock-multiline)
   font-lock-extend-region-multiline()
   font-lock-default-fontify-region(0 0 nil)
   font-lock-fontify-region(0 0)
   font-lock-after-change-function(1 11 0)
   compilation-start("make -k " nil)
   compile("make -k " nil)
   call-interactively(compile t nil)
   execute-extended-command(nil)
   call-interactively(execute-extended-command nil nil)


	Jan D.


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
/opt/src/emacs-cvs/etc/DEBUG for instructions.


In GNU Emacs 23.0.60.1 (i686-pc-linux-gnu, GTK+ Version 2.12.5)
  of 2008-04-08 on dentan.operax.com
Windowing system distributor `The X.Org Foundation', version 11.0.10300000
configured using `configure  '--prefix=/opt/emacs' '--verbose' 
'--enable-asserts' '--enable-font-backend' '--with-x-toolkit=gtk' 'CFLAGS=-g''

Important settings:
   value of $LC_ALL: nil
   value of $LC_COLLATE: C
   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: sv_SE.UTF-8
   value of $XMODIFIERS: @im=none
   locale-coding-system: utf-8-unix
   default-enable-multibyte-characters: t

Major mode: C/l

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-compression-mode: t
   line-number-mode: t
   transient-mark-mode: t
   abbrev-mode: t

Recent input:
<help-echo> M-x c o m p i l e <return> <return> <help-echo>
C-x C-f s r c <tab> g t <tab> c <return> M-x M-p <return>
<return> <help-echo> <help-echo> M-x r e p o <tab>
r <tab> <return>

Recent messages:
For information about GNU Emacs and the GNU system, type C-h C-a.
(No files need saving)
Compilation finished
Loading vc-cvs...done
(No files need saving)
font-lock-extend-region-multiline: Args out of range: 0, 0
Making completion list...

Re: 23.0.60; M-x compile gives args out of range 0, 0

[Stefan Monnier]

> When doing M-x compile I get:
> font-lock-extend-region-multiline: Args out of range: 0, 0

Does the patch below make it disappear, by any chance?


        Stefan


=== modified file 'lisp/progmodes/cc-mode.el'
--- lisp/progmodes/cc-mode.el	2008-04-07 16:37:22 +0000
+++ lisp/progmodes/cc-mode.el	2008-04-08 15:01:47 +0000
@@ -601,7 +601,7 @@
   (add-hook 'before-change-functions 'c-before-change nil t)
   (make-local-hook 'after-change-functions)
   (add-hook 'after-change-functions 'c-after-change nil t)
-  (setq font-lock-extend-after-change-region-function
+  (set (make-local-variable 'font-lock-extend-after-change-region-function)
  	'c-extend-after-change-region))	; Currently (2008-04), only used by AWK.
 
 (defun c-setup-doc-comment-style ()

Re: 23.0.60; M-x compile gives args out of range 0, 0

[Jan Djärv]

Stefan Monnier skrev:
>> When doing M-x compile I get:
>> font-lock-extend-region-multiline: Args out of range: 0, 0
> 
> Does the patch below make it disappear, by any chance?
> 
> 

Aha, it is cc-mode.  No wonder it looked random, I don't always load a c-file
before M-x compile.

The patch does indeed fix the problem.

	Jan D.

>         Stefan
> 
> 
> === modified file 'lisp/progmodes/cc-mode.el'
> --- lisp/progmodes/cc-mode.el	2008-04-07 16:37:22 +0000
> +++ lisp/progmodes/cc-mode.el	2008-04-08 15:01:47 +0000
> @@ -601,7 +601,7 @@
>    (add-hook 'before-change-functions 'c-before-change nil t)
>    (make-local-hook 'after-change-functions)
>    (add-hook 'after-change-functions 'c-after-change nil t)
> -  (setq font-lock-extend-after-change-region-function
> +  (set (make-local-variable 'font-lock-extend-after-change-region-function)
>   	'c-extend-after-change-region))	; Currently (2008-04), only used by AWK.
>  
>  (defun c-setup-doc-comment-style ()
> 
> 
> 

Re: 23.0.60; M-x compile gives args out of range 0, 0

[Alan Mackenzie]

Hi, everybody!

On Tue, Apr 08, 2008 at 11:03:02AM -0400, Stefan Monnier wrote:
> > When doing M-x compile I get:
> > font-lock-extend-region-multiline: Args out of range: 0, 0
> 
> Does the patch below make it disappear, by any chance?
> 
> 
>         Stefan
> 
> 
> === modified file 'lisp/progmodes/cc-mode.el'
> --- lisp/progmodes/cc-mode.el	2008-04-07 16:37:22 +0000
> +++ lisp/progmodes/cc-mode.el	2008-04-08 15:01:47 +0000

With respect, that's the wrong place to patch.  It is liable to catch
out anybody else who uses font-lock-extend-after-change-region-function
in the future.  This variable should be made buffer-local once and for
all, like this:


Index: font-lock.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/font-lock.el,v
retrieving revision 1.333
diff -c -r1.333 font-lock.el
*** font-lock.el	14 Mar 2008 16:15:47 -0000	1.333
--- font-lock.el	8 Apr 2008 17:01:34 -0000
***************
*** 996,1002 ****
  (defvar font-lock-extend-after-change-region-function nil
    "A function that determines the region to refontify after a change.
  
! This variable is either nil, or is a function that determines the
  region to refontify after a change.
  It is usually set by the major mode via `font-lock-defaults'.
  Font-lock calls this function after each buffer change.
--- 996,1002 ----
  (defvar font-lock-extend-after-change-region-function nil
    "A function that determines the region to refontify after a change.
  
! This buffer-local variable is either nil, or is a function that determines the
  region to refontify after a change.
  It is usually set by the major mode via `font-lock-defaults'.
  Font-lock calls this function after each buffer change.
***************
*** 1007,1012 ****
--- 1007,1013 ----
  \(which directs the caller to fontify a default region).
  This function should preserve the match-data.
  The region it returns may start or end in the middle of a line.")
+ (make-variable-buffer-local 'font-lock-extend-after-change-region-function)
  
  (defun font-lock-fontify-buffer ()
    "Fontify the current buffer the way the function `font-lock-mode' would."


-- 
Alan Mackenzie (Nuremberg, Germany).

Re: 23.0.60; M-x compile gives args out of range 0, 0

[Stefan Monnier]

>> > When doing M-x compile I get:
>> > font-lock-extend-region-multiline: Args out of range: 0, 0
>> 
>> Does the patch below make it disappear, by any chance?
>> 
>> 
>> Stefan
>> 
>> 
>> === modified file 'lisp/progmodes/cc-mode.el'
>> --- lisp/progmodes/cc-mode.el	2008-04-07 16:37:22 +0000
>> +++ lisp/progmodes/cc-mode.el	2008-04-08 15:01:47 +0000

> With respect, that's the wrong place to patch.  It is liable to catch
> out anybody else who uses font-lock-extend-after-change-region-function
> in the future.  This variable should be made buffer-local once and for
> all, like this:

I disagree.  A major mode should basically *never* change globally
a defvar, except maybe for its own vars (plus a few exceptions, of
course).  I.e. `setq' should only ever be used on let-bound variables.
If you happen to know that the variable is "automatically buffer-local",
it's OK to use `setq', but it's always safer to use
`make-local-variable', especially because unless the package that does
the defvar (and make-variable-buffer-local) is preloaded you may end up
calling `setq' before make-variable-buffer-local and get into trouble.

I'm not opposed to make-variable-buffer-local, but the bug was clearly
in CC-mode, because a major mode should by default use
(set (make-local-variable ...) ...) rather than setq.

BTW, why doesn't CC-mode set those vars via font-lock-defaults, as the
author of font-lock intended?


        Stefan

Re: 23.0.60; M-x compile gives args out of range 0, 0

[Alan Mackenzie]

Hi, Stefan!

On Tue, Apr 08, 2008 at 02:05:07PM -0400, Stefan Monnier wrote:
> >> > When doing M-x compile I get:
> >> > font-lock-extend-region-multiline: Args out of range: 0, 0

> >> Does the patch below make it disappear, by any chance?

> >> Stefan


> >> === modified file 'lisp/progmodes/cc-mode.el'
> >> --- lisp/progmodes/cc-mode.el	2008-04-07 16:37:22 +0000
> >> +++ lisp/progmodes/cc-mode.el	2008-04-08 15:01:47 +0000

> > With respect, that's the wrong place to patch.  It is liable to catch
> > out anybody else who uses
> > font-lock-extend-after-change-region-function in the future.  This
> > variable should be made buffer-local once and for all, like this:

> I disagree.  A major mode should basically *never* change globally
> a defvar, except maybe for its own vars (plus a few exceptions, of
> course).

Well, that's a bit philosophical.  In practice, major mode maintainers
_will_ be setting such variables without explicitly localising them,
causing just that little bit extra in debugging, stress, curse words, and
so on.

There's a category of variables which are _essentially_ buffer local, and
nobody ever explicitly localises them first - things like major-mode,
mark-ring, foo-minor-mode, buffer-undo-list, .....  fl-e-ac-rf is in this
category.

There are several other variables in font-lock.el which are explicity
buffer-localised, amongst them the similar variable
font-lock-extend-region-functions.  So it would promote consistency if we
did the same with fl-extend-ac-region-function.

[ .... ]

> I'm not opposed to make-variable-buffer-local, but the bug was clearly
> in CC-mode, because a major mode should by default use
> (set (make-local-variable ...) ...) rather than setq.

I don't think it's clear at all.  The same "bug", if such it be, exists
in font-lock.el itself, L1452:

        (setq font-lock-syntactically-fontified end))

> BTW, why doesn't CC-mode set those vars via font-lock-defaults, as the
> author of font-lock intended?

For portability's sake.  font-lock-defaults in XEmacs has a different
format.  In particular, it lacks the "other-vars" bit at the end.  It's
less hassle just not to use it.

>         Stefan

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: 23.0.60; M-x compile gives args out of range 0, 0

[Stefan Monnier]

>> I disagree.  A major mode should basically *never* change globally
>> a defvar, except maybe for its own vars (plus a few exceptions, of
>> course).

> Well, that's a bit philosophical.  In practice, major mode maintainers
> _will_ be setting such variables without explicitly localising them,
> causing just that little bit extra in debugging, stress, curse words,
> and so on.

It's not philosophical.  It's simply needed so as to avoid undesired
interference between unrelated buffers.  I know it's a common bug.

> There's a category of variables which are _essentially_ buffer local, and
> nobody ever explicitly localises them first - things like major-mode,
> mark-ring, foo-minor-mode, buffer-undo-list, .....

It's hard to clearly define which ones are "essentially" buffer local
and which ones aren't.

> fl-e-ac-rf is in this category.

In my mind, it's definitely not in the same category as `major-mode' or
`buffer-undo-list'.  But it doesn't matter anyway.

> There are several other variables in font-lock.el which are explicity
> buffer-localised, amongst them the similar variable
> font-lock-extend-region-functions.  So it would promote consistency if we
> did the same with fl-extend-ac-region-function.

As I said:

>> I'm not opposed to make-variable-buffer-local,

So feel free to apply your patch (except for the part that changes the
docstring, since C-h v will take care of warning the user that the
variable is automatically made buffer-local).

>> but the bug was clearly in CC-mode, because a major mode should by
>> default use (set (make-local-variable ...) ...) rather than setq.

> I don't think it's clear at all.  The same "bug", if such it be, exists
> in font-lock.el itself, L1452:

>         (setq font-lock-syntactically-fontified end))

As I said "If you happen to know that the variable is "automatically
buffer-local", it's OK to use `setq'".

>> BTW, why doesn't CC-mode set those vars via font-lock-defaults, as the
>> author of font-lock intended?

> For portability's sake.  font-lock-defaults in XEmacs has a different
> format.  In particular, it lacks the "other-vars" bit at the end.  It's
> less hassle just not to use it.

Duh!


        Stefan

Re: 23.0.60; M-x compile gives args out of range 0, 0

[Alan Mackenzie]

Morning, Stefan!

On Tue, Apr 08, 2008 at 04:22:23PM -0400, Stefan Monnier wrote:
> > There's a category of variables which are _essentially_ buffer local,
> > and nobody ever explicitly localises them first - things like
> > major-mode, mark-ring, foo-minor-mode, buffer-undo-list, .....

> It's hard to clearly define which ones are "essentially" buffer local
> and which ones aren't.

I'm not so sure about that.  Anything set by a major mode is almost
certainly such a variable.  font-lock-keywords, for example.  :-)  But
this is going towards the philosophical again.

[ .... ]

> So feel free to apply your patch (except for the part that changes the
> docstring, since C-h v will take care of warning the user that the
> variable is automatically made buffer-local).

DONE.

[ .... ]

>         Stefan

-- 
Alan Mackenzie (Nuremberg, Germany).