[Bug] in newcomment.el

[Bug] in newcomment.el

[Thorsten Jolitz]

Hi List, 

newcomment.el uses (line 477)

,------------------------------------
| (if (looking-at comment-start-skip)
`------------------------------------

in function

,--------------------------------------------------------
| (defun comment-search-forward (limit &optional noerror)
`--------------------------------------------------------

what gives an error if `comment-start-skip' is nil, as in sql-mode. 

E.g. this small file (sql-test.sql) in a buffer in sql-mode

,----------------------
| -- * data load
| -- ** data for table1
| drop table1;
`----------------------

with point at (point-min), causes an error when this command is applied:

,----------------------------------
| M-: (comment-search-forward 15 t)
`----------------------------------

,----------------------------------------------------------------
| Debugger entered--Lisp error: (wrong-type-argument stringp nil)
|   looking-at(nil)
|   comment-search-forward(15 t)
|   eval((comment-search-forward 15 t) nil)
|   eval-expression((comment-search-forward 15 t) nil)
|   call-interactively(eval-expression nil nil)
`----------------------------------------------------------------

because 'C-h v comment-start-skip' gives

,-------------------------------------------------------------
| comment-start-skip is a variable defined in `newcomment.el'.
| Its value is nil
`-------------------------------------------------------------

-- 
cheers,
Thorsten

RE: [Bug] in newcomment.el

[Drew Adams]

If you think you have found a bug, use `M-x report-emacs-bug RET'.

Re: [Bug] in newcomment.el

[Thorsten Jolitz]

Drew Adams <drew.adams@oracle.com> writes:

> If you think you have found a bug, use `M-x report-emacs-bug RET'.

ok, done. 

-- 
cheers,
Thorsten

Re: [Bug] in newcomment.el

[Glenn Morris]

Thorsten Jolitz wrote:

> ok, done. 

A shame neither of the people prompting you to report bugs suggested
searching for existing reports first.

http://debbugs.gnu.org/cgi/bugreport.cgi?bug=14376

With some familiar commenters.

Re: [Bug] in newcomment.el

[Thorsten Jolitz]

Glenn Morris <rgm@gnu.org> writes:

> Thorsten Jolitz wrote:

> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=14376

indeed, there it is ...

One question:

,---------------------------------------------------------
| You failed to call comment-normalize-vars before calling
| comment-search-forward.
`---------------------------------------------------------

do I have to call `comment-normalize-vars' each time I call a
(comment-...) function, or is 

,-----------------------------------------------------
| (add-hook 'my-library-hook 'comment-normalize-vars')
`-----------------------------------------------------

enough?

-- 
cheers,
Thorsten

RE: [Bug] in newcomment.el

[Drew Adams]

> A shame neither of the people prompting you to report bugs suggested
> searching for existing reports first.

A minor bother, perhaps, but no shame.

The important thing is to get the (doc) bug fixed.  If it takes another
bug report (and apparently it did in this case), so be it.  But yes,
sure, if there already is a report (or two), then presumably no additional
reporting is needed. ;-)

> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=14376
> 
> With some familiar commenters.

Including this clairvoyant comment from me ;-):

>> FWIW, a comment in the source code is hardly documentation for the
>> functions concerned.  Doc strings...
>>
>> See also bug #12583...  This will probably keep coming up
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> until there is some mention of it in the actual doc.

Thanks for documenting this now.

This thread has additional information/guidance about this issue:
http://debbugs.gnu.org/cgi/bugreport.cgi?bug=12583

Re: [Bug] in newcomment.el

[Andreas Röhler]

Am 18.06.2013 19:08, schrieb Thorsten Jolitz:
>
> Hi List,
>
> newcomment.el uses (line 477)
>
> ,------------------------------------
> | (if (looking-at comment-start-skip)
> `------------------------------------
>
> in function
>
> ,--------------------------------------------------------
> | (defun comment-search-forward (limit &optional noerror)
> `--------------------------------------------------------
>
> what gives an error if `comment-start-skip' is nil, as in sql-mode.
>
> E.g. this small file (sql-test.sql) in a buffer in sql-mode
>
> ,----------------------
> | -- * data load
> | -- ** data for table1
> | drop table1;
> `----------------------
>
> with point at (point-min), causes an error when this command is applied:
>
> ,----------------------------------
> | M-: (comment-search-forward 15 t)
> `----------------------------------
>
> ,----------------------------------------------------------------
> | Debugger entered--Lisp error: (wrong-type-argument stringp nil)
> |   looking-at(nil)
> |   comment-search-forward(15 t)
> |   eval((comment-search-forward 15 t) nil)
> |   eval-expression((comment-search-forward 15 t) nil)
> |   call-interactively(eval-expression nil nil)
> `----------------------------------------------------------------
>
> because 'C-h v comment-start-skip' gives
>
> ,-------------------------------------------------------------
> | comment-start-skip is a variable defined in `newcomment.el'.
> | Its value is nil
> `-------------------------------------------------------------
>

Looking forward for a bug-report too. `comment-start' should do a better job at this place.

Andreas