bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Chong Yidong]

With latest trunk, I do this:

1. M-x debbugs-gnu-search RET [some search string] RET
2. [Move point to one of the entries in the *Emacs Bugs* buffer]
3. RET
4. In the ephemeral Gnus buffer, type `r' to reply to a message.
5. In the *unsent reply* buffer, type `C-c C-c' to send.

Debugger entered--Lisp error: (error "No such function: nnspool-request-create-group")
  signal(error ("No such function: nnspool-request-create-group"))
  error("No such function: %s" nnspool-request-create-group)
  gnus-get-function((nnspool "" (nnspool-active-file "/dev/null")) request-create-group)
  gnus-request-create-group("sent.2012-07" (nnspool "" (nnspool-active-file "/dev/null")))
  gnus-inews-do-gcc()
  run-hooks(message-sent-hook)
  message-send(nil)
  message-send-and-exit(nil)

The message still gets delivered, so only the gcc seems to be failing.
I guess something in Gnus has changed.

Note that any fix committed to the debbug-gnus package should continue
to work with Emacs 24.1.


In GNU Emacs 24.1.50.10 (x86_64-unknown-linux-gnu, GTK+ Version 3.4.2)
 of 2012-07-14 on ulysses
Bzr revision: 109085 cyd@gnu.org-20120714021907-0txw01oym3aoqrwt

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Michael Albinus]

Chong Yidong <cyd@gnu.org> writes:

package 11941 emacs,gnus
thanks

> With latest trunk, I do this:
>
> 1. M-x debbugs-gnu-search RET [some search string] RET
> 2. [Move point to one of the entries in the *Emacs Bugs* buffer]
> 3. RET
> 4. In the ephemeral Gnus buffer, type `r' to reply to a message.
> 5. In the *unsent reply* buffer, type `C-c C-c' to send.
>
> Debugger entered--Lisp error: (error "No such function: nnspool-request-create-group")
>   signal(error ("No such function: nnspool-request-create-group"))
>   error("No such function: %s" nnspool-request-create-group)
>   gnus-get-function((nnspool "" (nnspool-active-file "/dev/null")) request-create-group)
>   gnus-request-create-group("sent.2012-07" (nnspool "" (nnspool-active-file "/dev/null")))
>   gnus-inews-do-gcc()
>   run-hooks(message-sent-hook)
>   message-send(nil)
>   message-send-and-exit(nil)
>
> The message still gets delivered, so only the gcc seems to be failing.
> I guess something in Gnus has changed.

This does not look like a debbugs-gnu bug. I suspect rather a gnus
problem.

I've added the "gnus" package to the report. This might increase the
visibility for the gnus maintainers :-)

Best regards, Miuchael.

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Glenn Morris]

Michael Albinus wrote:

> package 11941 emacs,gnus
> thanks

You want "reassign" not "package", and must send such commands to the
control server. (And it only affects bug comments from that point
onwards.)

> This does not look like a debbugs-gnu bug. I suspect rather a gnus
> problem.
>
> I've added the "gnus" package to the report. This might increase the
> visibility for the gnus maintainers :-)

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Michael Albinus]

Glenn Morris <rgm@gnu.org> writes:

>> package 11941 emacs,gnus
>> thanks
>
> You want "reassign" not "package", and must send such commands to the
> control server. (And it only affects bug comments from that point
> onwards.)

Thanks for fixing this.

I've used blindly the control message template what I've got by accessing
<http://debbugs.gnu.org/11941;mboxmaint=yes;mboxstat=yes>. This proposes

package 11941 emacs,gnus

which seems to be wrong. Could you fix it on the server?

Sending to the wrong address is my own error, I haven't found an
excuse. Sorry for that.

Best regards, Michael.

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Lars Ingebrigtsen]

Chong Yidong <cyd@gnu.org> writes:

>   gnus-get-function((nnspool "" (nnspool-active-file "/dev/null")) request-create-group)
>   gnus-request-create-group("sent.2012-07" (nnspool "" (nnspool-active-file "/dev/null")))
>   gnus-inews-do-gcc()

Looking at the code, it seems likely that

(gnus-inews-group-method "sent.2012-07")

returns nnspool instead of the real backend you wanted to use, for some
reason or other.  Is that correct?

Or perhaps it only returns that if the group doesn't exist at all.  What
does the following return?

(gnus-inews-group-method "sent.2012-07.no-such-group")

-- 
(domestic pets only, the antidote for overdose, milk.)
  http://lars.ingebrigtsen.no  *  Lars Magne Ingebrigtsen

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Lars Ingebrigtsen]

Hm.  Come to think of it, that Gcc header should have been something
like

nnfolder+archive:sent.2012w36

and not just the shorter form.  What does your Gcc header look like when
responding from a debbugs group?

-- 
(domestic pets only, the antidote for overdose, milk.)
  http://lars.ingebrigtsen.no  *  Lars Magne Ingebrigtsen

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Chong Yidong]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Hm.  Come to think of it, that Gcc header should have been something
> like
>
> nnfolder+archive:sent.2012w36
>
> and not just the shorter form.  What does your Gcc header look like when
> responding from a debbugs group?

Gcc: sent.2012-09

> What does the following return?
>
> (gnus-inews-group-method "sent.2012-07.no-such-group")

(nnspool "" (nnspool-active-file "/dev/null"))

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Lars Ingebrigtsen]

Chong Yidong <cyd@gnu.org> writes:

>> and not just the shorter form.  What does your Gcc header look like when
>> responding from a debbugs group?
>
> Gcc: sent.2012-09

Right.  That's the error.  Does your Gcc header look like this in all
groups?  It should look something like "nnfolder+archive:sent.2012-09".

-- 
(domestic pets only, the antidote for overdose, milk.)
  http://lars.ingebrigtsen.no  *  Lars Magne Ingebrigtsen

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Chong Yidong]

Lars Ingebrigtsen <larsi@gnus.org> writes:

>>> and not just the shorter form.  What does your Gcc header look like when
>>> responding from a debbugs group?
>>
>> Gcc: sent.2012-09
>
> Right.  That's the error.  Does your Gcc header look like this in all
> groups?  It should look something like "nnfolder+archive:sent.2012-09".

When I reply to you via Gnus, the Gcc header is

  Gcc: nnfolder+archive:sent.2012-09

Similarly, on Emacs 24.2 when I respond from a debbugs group, the Gcc
header is also

  Gcc: nnfolder+archive:sent.2012-09

On trunk, when I respond from a debbugs group, the Gcc header is the
faulty one given above.

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Lars Ingebrigtsen]

Chong Yidong <cyd@gnu.org> writes:

> When I reply to you via Gnus, the Gcc header is
>
>   Gcc: nnfolder+archive:sent.2012-09
>
> Similarly, on Emacs 24.2 when I respond from a debbugs group, the Gcc
> header is also
>
>   Gcc: nnfolder+archive:sent.2012-09
>
> On trunk, when I respond from a debbugs group, the Gcc header is the
> faulty one given above.

The Gcc header is inserted by `gnus-inews-insert-gcc'.  I've traced
through the function, and the following form is the one evaluated to
insert the group name.

(gnus-group-prefixed-name "sent.2012-09" gnus-message-archive-method)

What does it eval to for you?

-- 
(domestic pets only, the antidote for overdose, milk.)
  http://lars.ingebrigtsen.no  *  Lars Magne Ingebrigtsen

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Chong Yidong]

Lars Ingebrigtsen <larsi@gnus.org> writes:

>> Similarly, on Emacs 24.2 when I respond from a debbugs group, the Gcc
>> header is also
>>
>>   Gcc: nnfolder+archive:sent.2012-09
>>
>> On trunk, when I respond from a debbugs group, the Gcc header is the
>> faulty one given above.
>
> The Gcc header is inserted by `gnus-inews-insert-gcc'.  I've traced
> through the function, and the following form is the one evaluated to
> insert the group name.
>
> (gnus-group-prefixed-name "sent.2012-09" gnus-message-archive-method)
>
> What does it eval to for you?

Evaluating this in the (Message mode) reply buffer, evaluating the above
gives "sent.2012-09".  Evaluating `gnus-message-archive-method' gives
"archive".  Looking through the `gnus-group-prefixed-name' function, I
see that this happens because

(gnus-server-to-method "archive")

returns nil.  Perhaps because some part of Gnus has not been
initialized?  (I did not run M-x gnus before using debbug-gnus.)

Am I to understand that you can't reproduce this?  It is 100%
reproducible for me.

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Lars Ingebrigtsen]

Chong Yidong <cyd@gnu.org> writes:

> (gnus-server-to-method "archive")
>
> returns nil.  Perhaps because some part of Gnus has not been
> initialized?  (I did not run M-x gnus before using debbug-gnus.)

Oh, I see.  I somehow missed that Gnus wasn't running.

I'm able to reproduce this now.  However, I'm not sure how to fix this.
Even if the right Gcc line were inserted, no message would be stored on
the backend, because Gnus sets up all that stuff during startup.  For
instance, if we're storing on IMAP (which is pretty common these days),
Gnus kinda really has to go through the startup process to find out how
to contact the IMAP server.

So I think the right fix here is to not insert a Gcc header at all if
Gnus isn't running.

-- 
(domestic pets only, the antidote for overdose, milk.)
  http://lars.ingebrigtsen.no  *  Lars Magne Ingebrigtsen

bug#11941: 24.1.50; Error in debbugs-gnu with gnus-inews-do-gcc

[Lars Magne Ingebrigtsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> So I think the right fix here is to not insert a Gcc header at all if
> Gnus isn't running.

I've now done this.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/