Re: fortran-fill-paragraph fails

Re: fortran-fill-paragraph fails

[Michaël Cadilhac]

[-- Attachment #1.1.1: Type: text/plain, Size: 447 bytes --]

"Roland Winkler" <Roland.Winkler@physik.uni-erlangen.de> writes:

> Start a fresh emacs --no-init-file. Load file foo.f (fortran-mode):
> cat foo.f <<EOF
> C     This is a fortran comment
>       CALL FOO
> EOF
>
> On line 1 execute fill-paragraph, which will run
> fortran-fill-paragraph. This gives me
>
> C     This is a fortran comment ALL FOO

Thanks for reporting this bug.

Does the following fix it ? (And is it an acceptable solution ?)


[-- Attachment #1.1.2: fill.patch --]
[-- Type: text/x-patch, Size: 1579 bytes --]

Index: lisp/textmodes/fill.el
===================================================================
RCS file: /sources/emacs/emacs/lisp/textmodes/fill.el,v
retrieving revision 1.192
diff -c -r1.192 fill.el
*** lisp/textmodes/fill.el	6 Nov 2006 16:01:53 -0000	1.192
--- lisp/textmodes/fill.el	24 Dec 2006 00:12:07 -0000
***************
*** 838,844 ****
  	     (commark
  	      (comment-string-strip (buffer-substring comstart comin) nil t))
  	     (comment-re
! 	      (if (string-match comment-start-skip (concat commark "a"))
  		  (concat "[ \t]*" (regexp-quote commark)
  			  ;; Make sure we only match comments that use
  			  ;; the exact same comment marker.
--- 838,844 ----
  	     (commark
  	      (comment-string-strip (buffer-substring comstart comin) nil t))
  	     (comment-re
! 	      (if (string-match comment-start-skip (concat "\0" commark "a"))
  		  (concat "[ \t]*" (regexp-quote commark)
  			  ;; Make sure we only match comments that use
  			  ;; the exact same comment marker.
Index: lisp/ChangeLog
===================================================================
RCS file: /sources/emacs/emacs/lisp/ChangeLog,v
retrieving revision 1.10475
diff -C0 -r1.10475 ChangeLog
*** lisp/ChangeLog	23 Dec 2006 19:06:59 -0000	1.10475
--- lisp/ChangeLog	24 Dec 2006 00:13:34 -0000
***************
*** 0 ****
--- 1,5 ----
+ 2006-12-24  Michaël Cadilhac  <michael.cadilhac@lrde.org>
+ 
+ 	* textmodes/fill.el (fill-comment-paragraph): Prevent the use of
+ 	an optimized comment regexp if `comment-start-skip' uses a ^.
+ 

[-- Attachment #1.1.3: Type: text/plain, Size: 403 bytes --]


-- 
/!\ My mail address has changed, please update your files accordingly.
 |      Michaël `Micha' Cadilhac     |  C'est véritablement un scandale       |
 |         Epita/LRDE Promo 2007     |    et probablement une contrepéterie.  |
 |  http://michael.cadilhac.name     |          -- P. Desproges               |
 `--JID: michael.cadilhac@gmail.com--'                                   -  --'

[-- Attachment #1.2: Type: application/pgp-signature, Size: 188 bytes --]

[-- Attachment #2: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: fortran-fill-paragraph fails

[Roland Winkler]

On Sun Dec 24 2006 Michaël Cadilhac wrote:
> "Roland Winkler" <Roland.Winkler@physik.uni-erlangen.de> writes:
> > Start a fresh emacs --no-init-file. Load file foo.f (fortran-mode):
> > cat foo.f <<EOF
> > C     This is a fortran comment
> >       CALL FOO
> > EOF
> >
> > On line 1 execute fill-paragraph, which will run
> > fortran-fill-paragraph. This gives me
> >
> > C     This is a fortran comment ALL FOO
> 
> Thanks for reporting this bug.
> 
> Does the following fix it ? (And is it an acceptable solution ?)

Thanks a lot for the quick reply. For me, it works fine for the
above. Hopefully someone else can comment whether your patch is not
causing new problems which is unfortunately beyond my elisp
knowledge.

Roland

Re: fortran-fill-paragraph fails

[Michaël Cadilhac]

[-- Attachment #1.1: Type: text/plain, Size: 1516 bytes --]

michael@cadilhac.name (Michaël Cadilhac) writes:

> "Roland Winkler" <Roland.Winkler@physik.uni-erlangen.de> writes:
>
>> Start a fresh emacs --no-init-file. Load file foo.f (fortran-mode):
>> cat foo.f <<EOF
>> C     This is a fortran comment
>>       CALL FOO
>> EOF
>>
>> On line 1 execute fill-paragraph, which will run
>> fortran-fill-paragraph. This gives me
>>
>> C     This is a fortran comment ALL FOO
>
> Thanks for reporting this bug.
>
> Does the following fix it ? (And is it an acceptable solution ?)
>
> *** lisp/textmodes/fill.el	6 Nov 2006 16:01:53 -0000	1.192
> --- lisp/textmodes/fill.el	24 Dec 2006 00:12:07 -0000
> ***************
> *** 838,844 ****
>   	     (commark
>   	      (comment-string-strip (buffer-substring comstart comin) nil t))
>   	     (comment-re
> ! 	      (if (string-match comment-start-skip (concat commark "a"))
> --- 838,844 ----
>   	     (commark
>   	      (comment-string-strip (buffer-substring comstart comin) nil t))
>   	     (comment-re
> ! 	      (if (string-match comment-start-skip (concat "\0" commark "a"))

Installed.

-- 
/!\ My mail address has changed, please update your files accordingly.
 |      Michaël `Micha' Cadilhac     |  C'est véritablement un scandale       |
 |         Epita/LRDE Promo 2007     |    et probablement une contrepéterie.  |
 |  http://michael.cadilhac.name     |          -- P. Desproges               |
 `--JID: michael.cadilhac@gmail.com--'                                   -  --'

[-- Attachment #1.2: Type: application/pgp-signature, Size: 188 bytes --]

[-- Attachment #2: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: fortran-fill-paragraph fails

[Stefan Monnier]

> ***************
> *** 838,844 ****
>   	     (commark
>   	      (comment-string-strip (buffer-substring comstart comin) nil t))
>   	     (comment-re
> ! 	      (if (string-match comment-start-skip (concat commark "a"))
>   		  (concat "[ \t]*" (regexp-quote commark)
>   			  ;; Make sure we only match comments that use
>   			  ;; the exact same comment marker.
> --- 838,844 ----
>   	     (commark
>   	      (comment-string-strip (buffer-substring comstart comin) nil t))
>   	     (comment-re
> ! 	      (if (string-match comment-start-skip (concat "\0" commark "a"))
>   		  (concat "[ \t]*" (regexp-quote commark)
>   			  ;; Make sure we only match comments that use
>   			  ;; the exact same comment marker.

Please add a clear comment explaining the reason for this "\0".


        Stefan

Re: fortran-fill-paragraph fails

[Roland Winkler]

On Mon Dec 25 2006 Stefan Monnier wrote:
> >   	     (commark
> >   	      (comment-string-strip (buffer-substring comstart comin) nil t))
> >   	     (comment-re
> > ! 	      (if (string-match comment-start-skip (concat "\0" commark "a"))
> >   		  (concat "[ \t]*" (regexp-quote commark)
> >   			  ;; Make sure we only match comments that use
> >   			  ;; the exact same comment marker.
> 
> Please add a clear comment explaining the reason for this "\0".

I must admit that this patch goes much beyond my knowledge of elisp
hacking. Also, I cannot find anything in the elisp manual explaining
the meaning of "\0" in this context. If there is a general idea for
this, I'd much appreciate if it was documented in the elisp manual,
too. (Or did I look at the wrong spot?)

Roland

Re: fortran-fill-paragraph fails

[Stefan Monnier]

>> >   	     (commark
>> >   	      (comment-string-strip (buffer-substring comstart comin) nil t))
>> >   	     (comment-re
>> > ! 	      (if (string-match comment-start-skip (concat "\0" commark "a"))
>> >   		  (concat "[ \t]*" (regexp-quote commark)
>> >   			  ;; Make sure we only match comments that use
>> >   			  ;; the exact same comment marker.
>> 
>> Please add a clear comment explaining the reason for this "\0".

> I must admit that this patch goes much beyond my knowledge of elisp
> hacking. Also, I cannot find anything in the elisp manual explaining
> the meaning of "\0" in this context. If there is a general idea for
> this, I'd much appreciate if it was documented in the elisp manual,
> too. (Or did I look at the wrong spot?)

There's nothing specific to elisp there, and the \0 is just some
arbitrary char: a space would have probably worked just as well.


        Stefan

Re: fortran-fill-paragraph fails

[Michaël Cadilhac]

[-- Attachment #1.1: Type: text/plain, Size: 1455 bytes --]

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

>>> >   	     (commark
>>> >   	      (comment-string-strip (buffer-substring comstart comin) nil t))
>>> >   	     (comment-re
>>> > ! 	      (if (string-match comment-start-skip (concat "\0" commark "a"))
>>> >   		  (concat "[ \t]*" (regexp-quote commark)
>>> >   			  ;; Make sure we only match comments that use
>>> >   			  ;; the exact same comment marker.
>>> 
>>> Please add a clear comment explaining the reason for this "\0".
>
>> I must admit that this patch goes much beyond my knowledge of elisp
>> hacking. Also, I cannot find anything in the elisp manual explaining
>> the meaning of "\0" in this context. If there is a general idea for
>> this, I'd much appreciate if it was documented in the elisp manual,
>> too. (Or did I look at the wrong spot?)
>
> There's nothing specific to elisp there, and the \0 is just some
> arbitrary char: a space would have probably worked just as well.

... but it would have needed some thinking to get rid of the
« probably » ;-)

I've installed a comment for that.  Thanks for your help Stefan!

-- 
 |      Michaël `Micha' Cadilhac     |    Le second degré,                    |
 |         Epita/LRDE Promo 2007     |       c'est un peu                     |
 |  http://michael.cadilhac.name     |   le verlan sémantique.                |
 `--JID: michael.cadilhac@gmail.com--'                                   -  --'

[-- Attachment #1.2: Type: application/pgp-signature, Size: 188 bytes --]

[-- Attachment #2: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: fortran-fill-paragraph fails

[Richard Stallman]

    There's nothing specific to elisp there, and the \0 is just some
    arbitrary char: a space would have probably worked just as well.

What purpose does that character serve?  Is the aim to prevent the
regexp from ever matching, in practice?

Re: fortran-fill-paragraph fails

[Michaël Cadilhac]

[-- Attachment #1.1: Type: text/plain, Size: 1382 bytes --]

Richard Stallman <rms@gnu.org> writes:

>     There's nothing specific to elisp there, and the \0 is just some
>     arbitrary char: a space would have probably worked just as well.
>
> What purpose does that character serve?  Is the aim to prevent the
> regexp from ever matching, in practice?

No!  The aim of \0 is to prevent the match when the regexp is "^C",
for example, but also "^x?C" (no mode does that AFAIK, but it's just
a security) where x is any non regexp-special char except \0.

Suppose commark is `C', we will have :
(string-match "^C" (concat "\0" commark "a")) -> nil
(string-match "C" (concat "\0" commark "a")) -> 1
(string-match "^ ?C" (concat "\0" commark "a")) -> nil

If ` ' was used instead of "\0", the last one would have returned 0.

Of course, it doesn't prevent a regexp like "^.C", but we can suppose,
IMO, that no mode will do that.  And IMO, it would be overkill to do
something to prevent that. "\0" is just a more secure char in this
precise context, at no cost :-) (except this thread ;-))

-- 
 |      Michaël `Micha' Cadilhac     |  In a World without Walls and Fences,  |
 |         Epita/LRDE Promo 2007     |     who needs Windows and Gates?       |
 |  http://michael.cadilhac.name     |          -- Dino Esposito              |
 `--JID: michael.cadilhac@gmail.com--'                                   -  --'

[-- Attachment #1.2: Type: application/pgp-signature, Size: 188 bytes --]

[-- Attachment #2: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: fortran-fill-paragraph fails

[Roland Winkler]

On Fri Dec 29 2006 Stefan Monnier wrote:
> There's nothing specific to elisp there, and the \0 is just some
> arbitrary char: a space would have probably worked just as well.

Thank you, slowly I start to understand what happens. The char must
be "sufficiently arbitrary" to do the job. If "\0" is replaced by
"C" it doesn't solve the bug. It is rather unlikely that "\0" has
the special meaning in some other mode which "C" has in fortran-mode.

Roland

Re: fortran-fill-paragraph fails

[Richard Stallman]

    Suppose commark is `C', we will have :
    (string-match "^C" (concat "\0" commark "a")) -> nil
    (string-match "C" (concat "\0" commark "a")) -> 1
    (string-match "^ ?C" (concat "\0" commark "a")) -> nil

Why are these results correct?  What is the overall explanation for
the job this code is trying to do?

I find I can't understand these comments

	      ;; `commark' is surrounded with arbitrary text (`\0' and `a')
	      ;;  to make sure it can be used as an optimization of
	      ;; `comment-start-skip' in the middle of a line.  For example,
	      ;; `commark' can't be used with the "@c" in TeXinfo (hence
	      ;; the `a') or with the "C" at BOL in Fortran (hence the `\0').

As far as I know, COMMARK (which is how it should be written) refers
to some text copied out of the buffer.  What does it mean to say
whether that that text "can't be used with the `@c' in Texinfo"?

Is there anyone that actually understands that comment
and could rewrite it more clearly?

Re: fortran-fill-paragraph fails

[Stefan Monnier]

>     Suppose commark is `C', we will have :
>     (string-match "^C" (concat "\0" commark "a")) -> nil
>     (string-match "C" (concat "\0" commark "a")) -> 1
>     (string-match "^ ?C" (concat "\0" commark "a")) -> nil

> Why are these results correct?  What is the overall explanation for
> the job this code is trying to do?

> I find I can't understand these comments

> 	      ;; `commark' is surrounded with arbitrary text (`\0' and `a')
> 	      ;;  to make sure it can be used as an optimization of
> 	      ;; `comment-start-skip' in the middle of a line.  For example,
> 	      ;; `commark' can't be used with the "@c" in TeXinfo (hence
> 	      ;; the `a') or with the "C" at BOL in Fortran (hence the `\0').

> As far as I know, COMMARK (which is how it should be written) refers
> to some text copied out of the buffer.  What does it mean to say
> whether that that text "can't be used with the `@c' in Texinfo"?

> Is there anyone that actually understands that comment
> and could rewrite it more clearly?

It seems a bit hard to understand indeed.  What the code around there is
trying to do is to try and come up with a more precise regexp than
comment-start-skip, which should not match any comment-start but only the
exact comment starter used in these comment lines.

So if the comment starter used is ";;;" the regexp can just be something
like ";;;[^;]".  But that's only for "typical" comment markers.
For Texinfo's "@c" we can't just use "@c[^c]" because "@ca" is not a comment
starter.  And For Fortran we can't just use "C[^C]" because the "C" is only
a comment starter when it's at the beginning of a line.

We heuristically try to discover whether it's a "typical" comment starter by
matching (concat "\0" comstart "a") against comment-start-skip, which should
correctly distinguish those two special cases.


        Stefan

Re: fortran-fill-paragraph fails

[Richard Stallman]

    So if the comment starter used is ";;;" the regexp can just be something
    like ";;;[^;]".  But that's only for "typical" comment markers.
    For Texinfo's "@c" we can't just use "@c[^c]" because "@ca" is not a comment
    starter.  And For Fortran we can't just use "C[^C]" because the "C" is only
    a comment starter when it's at the beginning of a line.

    We heuristically try to discover whether it's a "typical" comment starter by
    matching (concat "\0" comstart "a") against comment-start-skip, which should
    correctly distinguish those two special cases.

Could you make the comments explain this?

Re: fortran-fill-paragraph fails

[Stefan Monnier]

>     So if the comment starter used is ";;;" the regexp can just be
>     something like ";;;[^;]".  But that's only for "typical" comment
>     markers.  For Texinfo's "@c" we can't just use "@c[^c]" because "@ca"
>     is not a comment starter.  And For Fortran we can't just use "C[^C]"
>     because the "C" is only a comment starter when it's at the beginning
>     of a line.

>     We heuristically try to discover whether it's a "typical" comment
>     starter by matching (concat "\0" comstart "a") against
>     comment-start-skip, which should correctly distinguish those two
>     special cases.

> Could you make the comments explain this?

Done,


        Stefan