bug on ognus 0.05 and emacs 21.2.50.1

bug on ognus 0.05 and emacs 21.2.50.1

[John Covici]

[-- Attachment #1: Type: text/plain, Size: 535 bytes --]

There are two related bugs, one is that in this version of emacs, I
cannot complete any mail nic name with an _ in it and when I took out
the _ I got the following back trace.

Debugger entered--Lisp error: (wrong-type-argument syntax-table-p nil)
  set-syntax-table(nil)
  mail-abbrev-complete-alias()
  call-interactively(mail-abbrev-complete-alias)

Oort Gnus v0.05
GNU Emacs 21.2.50.1 (i686-pc-linux-gnu, X toolkit)
 of 2002-03-31 on ccs.covici.com
200 ccs.covici.com InterNetNews server INN 2.3.3 (20010809 prerelease) ready






[-- Attachment #2: User settings --]
[-- Type: application/emacs-lisp, Size: 7931 bytes --]

[-- Attachment #3: Type: text/plain, Size: 56 bytes --]

-- 
         John Covici
         covici@ccs.covici.com

Re: bug on ognus 0.05 and emacs 21.2.50.1

[Richard Stallman]

    There are two related bugs, one is that in this version of emacs, I
    cannot complete any mail nic name with an _ in it and when I took out
    the _ I got the following back trace.

I changed that code last.  Can you tell me the precise test case?
I can debug it myself if the test case does not require Gnus.

Re: bug on ognus 0.05 and emacs 21.2.50.1

[John Covici]

Well it was an alias in my .mailrc file for instance I had a line like
alias joel_dejean "jfdejean@att.net"

which won't complete because I get a message 'can't complete dejean'
when I use <esc>-tab.  However, if I use the same alias and take out
the underscore it gives the back trace I sent earlier complaining
about the syntax table.

Is this helpful?

Thanks.


On Tue, 2 Apr 2002, Richard Stallman wrote:

>     There are two related bugs, one is that in this version of emacs, I
>     cannot complete any mail nic name with an _ in it and when I took out
>     the _ I got the following back trace.
>
> I changed that code last.  Can you tell me the precise test case?
> I can debug it myself if the test case does not require Gnus.
>

-- 
         John Covici
         covici@ccs.covici.com

Re: bug on ognus 0.05 and emacs 21.2.50.1

[ShengHuo ZHU]

Richard Stallman <rms@gnu.org> writes:

>     There are two related bugs, one is that in this version of emacs, I
>     cannot complete any mail nic name with an _ in it and when I took out
>     the _ I got the following back trace.
>
> I changed that code last.  Can you tell me the precise test case?
> I can debug it myself if the test case does not require Gnus.

Reproduce it in the scratch buffer:
  
  M-x load-library mailabbrev RET
  M-x mail-abbrev-complete-alias RET

I think, calling mail-abbrev-make-syntax-table in the very beginning
of mail-abbrev-complete-alias fixes the problem.

ShengHuo

Re: bug on ognus 0.05 and emacs 21.2.50.1

[Richard Stallman]

    I think, calling mail-abbrev-make-syntax-table in the very beginning
    of mail-abbrev-complete-alias fixes the problem.

Ok, I did that.  Thanks.  John, does this fix work?

Index: mailabbrev.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/mail/mailabbrev.el,v
retrieving revision 1.65
retrieving revision 1.66
diff -c -c -r1.65 -r1.66
*** mailabbrev.el	23 Mar 2002 16:09:59 -0000	1.65
--- mailabbrev.el	3 Apr 2002 15:33:06 -0000	1.66
***************
*** 548,553 ****
--- 548,554 ----
    "Perform completion on alias preceding point."
    ;; Based on lisp.el:lisp-complete-symbol
    (interactive)
+   (mail-abbrev-make-syntax-table)
    (let* ((end (point))
  	 (syntax-table (syntax-table))
  	 (beg (unwind-protect

Re: bug on ognus 0.05 and emacs 21.2.50.1

[John Covici]

It fixed the lisp error, but any alias with an _ still does not
work.  Did someone change the syntax table on us or is this the
desired behavior?

Also, when I was doing this I noticed that the make install didn't
compile newer versions of the mailabbrev.el and byte-opt.el, but did
copy them into the /usr/share/emacs/21.2.50.2 directory, but left the old elc
files there.  Is this what is supposed to happen?

on Wed, 3 Apr 2002 15:49:12 -0700 (MST) Richard Stallman <rms@gnu.org> wrote:

>     I think, calling mail-abbrev-make-syntax-table in the very beginning
>     of mail-abbrev-complete-alias fixes the problem.
>
> Ok, I did that.  Thanks.  John, does this fix work?
>
> Index: mailabbrev.el
> ===================================================================
> RCS file: /cvsroot/emacs/emacs/lisp/mail/mailabbrev.el,v
> retrieving revision 1.65
> retrieving revision 1.66
> diff -c -c -r1.65 -r1.66
> *** mailabbrev.el	23 Mar 2002 16:09:59 -0000	1.65
> --- mailabbrev.el	3 Apr 2002 15:33:06 -0000	1.66
> ***************
> *** 548,553 ****
> --- 548,554 ----
>     "Perform completion on alias preceding point."
>     ;; Based on lisp.el:lisp-complete-symbol
>     (interactive)
> +   (mail-abbrev-make-syntax-table)
>     (let* ((end (point))
>   	 (syntax-table (syntax-table))
>   	 (beg (unwind-protect

-- 
         John Covici
         covici@ccs.covici.com

Re: bug on ognus 0.05 and emacs 21.2.50.1

[Eli Zaretskii]

> From: John Covici <covici@ccs.covici.com>
> Date: Sat, 06 Apr 2002 00:54:41 -0500
> 
> Also, when I was doing this I noticed that the make install didn't
> compile newer versions of the mailabbrev.el and byte-opt.el, but did
> copy them into the /usr/share/emacs/21.2.50.2 directory, but left the old elc
> files there.  Is this what is supposed to happen?

AFAIK, "make" and "make install" never compile Lisp files.  If you
want to compile them, go to the lisp directory and type "make recompile".

Re: bug on ognus 0.05 and emacs 21.2.50.1

[Richard Stallman]

    It fixed the lisp error, but any alias with an _ still does not
    work.

Can you please send a precise self-contained test case of the current failure?

Re: bug on ognus 0.05 and emacs 21.2.50.1

[John Covici]

I sent it before, but I can send it again or just describe it.

alias joel_dejean "jfdejean@att.net"

If you try it it complains that it cannot complete dejean which seems
to mean that it is ignoring what is before the _ .  If I imit the _
it now works correctly.


On Sat, 6 Apr 2002, Richard Stallman wrote:

>     It fixed the lisp error, but any alias with an _ still does not
>     work.
>
> Can you please send a precise self-contained test case of the current failure?
>

-- 
         John Covici
         covici@ccs.covici.com