Bug sending messages unplugged

Bug sending messages unplugged

[Markus Dickebohm]

I get the following backtrace while sending queued messages ("J S")
with pgnus-0.53 (this is with XEmacs-20.4-non-mule):

Signaling: (wrong-type-argument integer-or-marker-p nil)
  lazy-shot-fontify-internal(#<buffer "*sent mail*<2>"> nil nil nil "")
  lazy-shot-lock-extent(#<destroyed extent> nil)
  lazy-shot-redisplay-function(#<destroyed extent>)

As far as I can see, the messages aren't sent although they disappear
from the "nndraft:queue"-group.

Markus
-- 
Uni Koeln * Physikalische Chemie II * Luxemburger Str. 116 * D-50939 Koeln
Tel: +49(0)221/470-3087  * Fax: -5144  *   mailto:M.Dickebohm@Uni-Koeln.DE

Re: Bug sending messages unplugged

[Colin Rafferty]

Markus Dickebohm writes:

> I get the following backtrace while sending queued messages ("J S")
> with pgnus-0.53 (this is with XEmacs-20.4-non-mule):

> Signaling: (wrong-type-argument integer-or-marker-p nil)
>   lazy-shot-fontify-internal(#<buffer "*sent mail*<2>"> nil nil nil "")
>   lazy-shot-lock-extent(#<destroyed extent> nil)
>   lazy-shot-redisplay-function(#<destroyed extent>)

This is very weird.  Is this the entire stack trace?  It seems a
little truncated.

-- 
Colin

Re: Bug sending messages unplugged

[Wes Hardaker]

>>>>> On 23 Nov 1998 14:58:32 -0500, Colin Rafferty <craffert@ms.com> said:

>> Signaling: (wrong-type-argument integer-or-marker-p nil)
>> lazy-shot-fontify-internal(#<buffer "*sent mail*<2>"> nil nil nil "")
>> lazy-shot-lock-extent(#<destroyed extent> nil)
>> lazy-shot-redisplay-function(#<destroyed extent>)

Colin> This is very weird.  Is this the entire stack trace?  It seems a
Colin> little truncated.

Yes it is, actually.  I get them too.  The odd thing is that I only
get them when sending news, not mail.

Anyway, I think what is happening is that gnus places the articles in
a new buffer (*mail* oddly enough, since its news) and then font lock
is swinging on them after that.  What I'm not sure of is if its
happening before or after the article gets sent.  I think its before
(which is bad) not after (which would be bad for me, because it means
I've sent multiple copies of messages to my newserver because of the
above bug).

I submitted a bug report to Lars, and he said its lazy-shot's fault
and I should submit the bug to that author instead.  I didn't want to
do that until determining the exact "fault" first, since I suspect
they'd say its was Gnus' fault.
-- 
"Ninjas aren't dangerous.  They're more afraid of you than you are of them."

Re: Bug sending messages unplugged

[Markus Dickebohm]

Colin Rafferty <craffert@ms.com> writes:

> This is very weird.  Is this the entire stack trace?  It seems a
> little truncated.

I reproduced the bug and now I get:

Signaling: (wrong-type-argument integer-or-marker-p nil)
  lazy-shot-fontify-internal(#<buffer "*sent mail*<3>"> nil nil nil "")
  lazy-shot-lock-extent(#<destroyed extent> nil)
  lazy-shot-redisplay-function(#<destroyed extent>)
  recursive-edit()
  byte-code("À	!ˆÂ ˆpÃÄ\x1d\x1e\x06\x1e\aÈ ˆ+ebˆÉ ˆ`ÊË!ˆÌyˆ`|ˆÍ ˆ\x0e\x0e@Ïs«›Ðcˆ\x0e\x0e@Ña«îÒÓÃ\"ˆÔÌ!ˆÕcˆÖyˆªÝ\x0e\x0e@×a«žØcˆ\x0e\x0eA@\x16\x19Ú\x0e\x19p\"ˆÛcˆÔÌ!ˆÜcˆÖyˆª¸\x0e\x0e@Ýa«ÞcˆÚ\x0e\x0eA@p\"ˆÛcˆª¡\x0e\x0e@Ãa«…ßcˆª•Ú\x0e\x0e@àa«…\x0e\x0eAª‚\x0e\x0ep\"ˆÛcˆáâ!ˆÃàÃ\x1e#\x1e\a\x1e$áâ!ˆå ˆ+à‡" [pop-to-buffer debugger-buffer erase-buffer t 50 print-length print-escape-newlines standard-output backtrace debugger-mode re-search-forward "\n[* ] debug(" 1 debugger-reenable debugger-args (lambda debug) "Entering:\n" debug backtrace-debug 3 delete-char ?\* 0 exit "Return value: " debugger-value prin1 ?\n ?\  error "Signaling: " "Beginning evaluation of function call form:\n" nil message "" buffer-read-only inhibit-trace recursive-edit] 3)
  debug(error (wrong-type-argument integer-or-marker-p nil))
  lazy-shot-fontify-internal(#<killed buffer> nil nil nil "")
  lazy-shot-lock-extent(#<destroyed extent> nil)
  lazy-shot-redisplay-function(#<destroyed extent>)

Wes Hardaker <wjhardaker@ucdavis.edu> writes:

> Yes it is, actually.  I get them too.  The odd thing is that I only
> get them when sending news, not mail.

This isn't the case here. I tried mail and news but in any case only
when sending MIME-Mails the error gets triggered.

My tests revealed another bug in respect to news postings:
 two Content-Type headers are generated when sending queued postings
 and my articles were rejected by the newsserver.

Markus
-- 
Uni Koeln * Physikalische Chemie II * Luxemburger Str. 116 * D-50939 Koeln
Tel: +49(0)221/470-3087  * Fax: -5144  *   mailto:M.Dickebohm@Uni-Koeln.DE

Re: Bug sending messages unplugged

[Wes Hardaker]

>>>>> On 24 Nov 1998 00:08:47 +0100, Markus Dickebohm <m.dickebohm@Uni-Koeln.DE> said:

Markus> My tests revealed another bug in respect to news postings: two
Markus> Content-Type headers are generated when sending queued
Markus> postings and my articles were rejected by the newsserver.

I saw this when re-posting the articles that died from the previous
error.  Mainly because I was submitting them with C-c C-c from the
mail buffer they died in, which caused the headers to get inserted a
second time.  Gnus should, though, remove the old headers before
recreating new ones.

-- 
"Ninjas aren't dangerous.  They're more afraid of you than you are of them."

Re: Bug sending messages unplugged

[Jan Vroonhof]

Wes Hardaker <wjhardaker@ucdavis.edu> writes:

> I submitted a bug report to Lars, and he said its lazy-shot's fault
> and I should submit the bug to that author instead.  I didn't want to
> do that until determining the exact "fault" first, since I suspect
> they'd say its was Gnus' fault.

It is too late now. I have seen it :-)

> >> lazy-shot-fontify-internal(#<buffer "*sent mail*<2>"> nil nil nil "")
> >> lazy-shot-lock-extent(#<destroyed extent> nil)
> >> lazy-shot-redisplay-function(#<destroyed extent>)

It is indeed a Lazy-shot bug. It is fixed in 21.0. Gnus is probably
copying parts of the buffer around into strings and lazy-shot does not
like that very much. However this is also a good indication Gnus is
using buffer-substring somewhere where it probably needs
buffer-substring-noproperties. 

===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs-packages/oa/edit-utils/lazy-shot.el,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- oa/edit-utils/lazy-shot.el  1998/01/14 06:36:17     1.1
+++ oa/edit-utils/lazy-shot.el  1998/04/18 14:31:24     1.2
@@ -157,7 +157,7 @@
 (defun lazy-shot-lock-extent (extent stealth)
   "Font-lock the EXTENT. Called from redisplay-trigger functions and
 stealth locking functions"
-   (when (extent-live-p extent)
+  (when (and (extent-live-p extent) (null (extent-detached-p extent)))
      (let ((start (extent-start-position extent))
           (end   (extent-end-position extent))
           (buffer (extent-object extent)))


> 
> Colin> This is very weird.  Is this the entire stack trace?  It seems a
> Colin> little truncated.
> 
> Yes it is, actually.  I get them too.  The odd thing is that I only
> get them when sending news, not mail.
> 
> Anyway, I think what is happening is that gnus places the articles in
> a new buffer (*mail* oddly enough, since its news) and then font lock
> is swinging on them after that.  What I'm not sure of is if its
> happening before or after the article gets sent.  I think its before
> (which is bad) not after (which would be bad for me, because it means
> I've sent multiple copies of messages to my newserver because of the
> above bug).
> 
> -- 
> "Ninjas aren't dangerous.  They're more afraid of you than you are of them."
> 

-- 
Jan Vroonhof                    http://www.math.ethz.ch/~vroonhof/
Mathematik,                               vroonhof @  math.ethz.ch
HG E16, ETH-Zentrum,                      Tel: +41-1-6325456/25154
Raemistrasse 101, CH-8092 Zuerich.              Fax: +41-1-6321085

Re: Bug sending messages unplugged

[Lars Magne Ingebrigtsen]

Colin Rafferty <craffert@ms.com> writes:

> > Signaling: (wrong-type-argument integer-or-marker-p nil)
> >   lazy-shot-fontify-internal(#<buffer "*sent mail*<2>"> nil nil nil "")
> >   lazy-shot-lock-extent(#<destroyed extent> nil)
> >   lazy-shot-redisplay-function(#<destroyed extent>)
> 
> This is very weird.  Is this the entire stack trace?  It seems a
> little truncated.

If lazy-shot is running off of after-change-hooks (or the like), you'd 
get a backtrace this short when it fails.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Bug sending messages unplugged

[Wes Hardaker]

>>>>> On 24 Nov 1998 11:03:56 +0100, Jan Vroonhof <vroonhof@math.ethz.ch> said:

Jan> It is too late now. I have seen it :-)

Cool! Thanks for the patch!

-- 
"Ninjas aren't dangerous.  They're more afraid of you than you are of them."