Problem displaying multipart/alternative messages

Problem displaying multipart/alternative messages

[Michael Piotrowski]

Hi,

Using the current git version of Gnus on XEmacs 21.4.22, I'm now getting
the error "Selecting deleted or non-existent buffer" when trying to view
a multipart/alternative message (as produced by, e.g., Apple Mail or
Microsoft Outlook):

#v+
Debugger entered--Lisp error: (error "Selecting deleted or non-existent buffer")
  mm-preferred-alternative-precedence(((#<buffer " *mm*<2>"> ("text/plain" ... ... ...) quoted-printable nil nil nil nil nil) ("multipart/related" (#<buffer " *mm*<4>"> ... quoted-printable nil nil nil nil nil) (#<buffer " *mm*<5>"> ... base64 nil ... nil nil "<169B2FFF-3C89-4C5A-BFAF-D4BF383302CD@local>"))))
  mm-preferred-alternative(((#<buffer " *mm*<2>"> ("text/plain" ... ... ...) quoted-printable nil nil nil nil nil) ("multipart/related" (#<buffer " *mm*<4>"> ... quoted-printable nil nil nil nil nil) (#<buffer " *mm*<5>"> ... base64 nil ... nil nil "<169B2FFF-3C89-4C5A-BFAF-D4BF383302CD@local>"))))
  gnus-mime-display-alternative(((#<buffer " *mm*<2>"> ("text/plain" ... ... ...) quoted-printable nil nil nil nil nil) ("multipart/related" (#<buffer " *mm*<4>"> ... quoted-printable nil nil nil nil nil) (#<buffer " *mm*<5>"> ... base64 nil ... nil nil "<169B2FFF-3C89-4C5A-BFAF-D4BF383302CD@local>"))) nil nil 1)
  gnus-mime-display-part(("multipart/alternative" (#<buffer " *mm*<2>"> ("text/plain" ... ... ...) quoted-printable nil nil nil nil nil) ("multipart/related" (#<buffer " *mm*<4>"> ... quoted-printable nil nil nil nil nil) (#<buffer " *mm*<5>"> ... base64 nil ... nil nil "<169B2FFF-3C89-4C5A-BFAF-D4BF383302CD@local>"))))
  gnus-display-mime()
  gnus-article-prepare-display()
  gnus-article-prepare(1860 nil)
  gnus-summary-display-article(1860 nil)
  gnus-summary-select-article(nil force)
  gnus-summary-show-article(nil)
  call-interactively(gnus-summary-show-article)
#v-

I have this in my .gnus:

#v+
;;; Discourage text/html alternatives
(setq mm-discouraged-alternatives '("text/html"))

;;; Don't render HTML, even if there's no alternative
;;; http://my.gnus.org/faq/html-nochunks/#id2491979
(setq mm-automatic-display (remove "text/html" mm-automatic-display))
#v-

So I first thought it may be a configuration incompatibility, but the
error also occurs with the default settings.

Any ideas?

Thanks in advance!

-- 
Dr.-Ing. Michael Piotrowski, M.A.                   <mxp@dynalabs.de>
Public key at <http://www.dynalabs.de/mxp/pubkey.txt> (ID 0x1614A044)

Re: Problem displaying multipart/alternative messages

[Michael Piotrowski]

On 2011-01-13, Michael Piotrowski <mxp@dynalabs.de> wrote:

> Using the current git version of Gnus on XEmacs 21.4.22, I'm now getting
> the error "Selecting deleted or non-existent buffer" when trying to view
> a multipart/alternative message (as produced by, e.g., Apple Mail or
> Microsoft Outlook):
>
> Debugger entered--Lisp error: (error "Selecting deleted or non-existent buffer")
>   mm-preferred-alternative-precedence(((#<buffer " *mm*<2>"> ("text/plain" ... ... ...) quoted-printable nil nil nil nil nil) ("multipart/related" (#<buffer " *mm*<4>"> ... quoted-printable nil nil nil nil nil) (#<buffer " *mm*<5>"> ... base64 nil ... nil nil "<169B2FFF-3C89-4C5A-BFAF-D4BF383302CD@local>"))))

[...]

> Any ideas?

The problem is the code after the comment "Remove empty parts" in
`mm-preferred-alternative-precedence'.  If I remove this code, it works
as before.

The messages I'm referring to in my original post are of type
"multipart/alternative", containing a "text/plain" and a
"multipart/mixed" part.  The latter part typicalle consists of a
"text/html" part and one or more attachments, e.g., "application/pdf".
When displaying such a message,

  (mm-handle-buffer handle)

returns the string "multipart/mixed" instead of a buffer, and
`with-current-buffer' then raises the error above.

Does really nobody else encounter this problem?

Best regards

-- 
Dr.-Ing. Michael Piotrowski, M.A.                   <mxp@dynalabs.de>
Public key at <http://www.dynalabs.de/mxp/pubkey.txt> (ID 0x1614A044)

Re: Problem displaying multipart/alternative messages

[Julien Danjou]

[-- Attachment #1: Type: text/plain, Size: 1483 bytes --]

On Thu, Jan 20 2011, Michael Piotrowski wrote:

> The problem is the code after the comment "Remove empty parts" in
> `mm-preferred-alternative-precedence'.  If I remove this code, it works
> as before.
>
> The messages I'm referring to in my original post are of type
> "multipart/alternative", containing a "text/plain" and a
> "multipart/mixed" part.  The latter part typicalle consists of a
> "text/html" part and one or more attachments, e.g., "application/pdf".
> When displaying such a message,
>
>   (mm-handle-buffer handle)
>
> returns the string "multipart/mixed" instead of a buffer, and
> `with-current-buffer' then raises the error above.
>
> Does really nobody else encounter this problem?

I encountered it this morning too.
The problem come from the commit

commit d9418f8743d26df6253803ccb2ff8cdc88e8934f
Author: Lars Magne Ingebrigtsen <larsi@quimbies.gnus.org>
Date:   Tue Jan 11 20:20:56 2011 +0100

    (mm-preferred-alternative-precedence): Discourage showing empty parts.


it introduces this regression because it does:

#+begin_src emacs-lisp
    (unless (with-current-buffer (mm-handle-buffer handle)
      …))
#+end_src

Which fails, because some of the handle is not like the other:
calling mm-handle-buffer on it returns a string, not a buffer. The
handle entry is not in the same format (with buffer in car) that the
other. But Lars must know that I think.

-- 
Julien Danjou
❱ http://julien.danjou.info

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

quoting Emacs Lisp (was: Problem displaying multipart/alternative messages)

[Ted Zlatanov]

On Thu, 20 Jan 2011 18:55:36 +0100 Julien Danjou <julien@danjou.info> wrote: 

JD> #+begin_src emacs-lisp
JD>     (unless (with-current-buffer (mm-handle-buffer handle)
JD>       …))
JD> #+end_src

Where are these markers from?  Are they standard in any way?

Ted

Re: quoting Emacs Lisp

[Julien Danjou]

[-- Attachment #1: Type: text/plain, Size: 540 bytes --]

On Thu, Jan 20 2011, Ted Zlatanov wrote:

> On Thu, 20 Jan 2011 18:55:36 +0100 Julien Danjou <julien@danjou.info> wrote: 
>
> JD> #+begin_src emacs-lisp
> JD>     (unless (with-current-buffer (mm-handle-buffer handle)
> JD>       …))
> JD> #+end_src
>
> Where are these markers from?  Are they standard in any way?

*cough* *cough*

I was just trying my recent pushed patches.[1]

[1]  http://julien.danjou.info/blog/2011.html#Code_fontification_with_Gnus_and_Org-mode

-- 
Julien Danjou
❱ http://julien.danjou.info

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: quoting Emacs Lisp

[Ted Zlatanov]

On Thu, 20 Jan 2011 22:05:55 +0100 Julien Danjou <julien@danjou.info> wrote: 

JD> On Thu, Jan 20 2011, Ted Zlatanov wrote:
>> On Thu, 20 Jan 2011 18:55:36 +0100 Julien Danjou <julien@danjou.info> wrote: 
>> 
JD> #+begin_src emacs-lisp
JD> (unless (with-current-buffer (mm-handle-buffer handle)
JD> …))
JD> #+end_src
>> 
>> Where are these markers from?  Are they standard in any way?

JD> *cough* *cough*

JD> I was just trying my recent pushed patches.[1]

JD> [1]  http://julien.danjou.info/blog/2011.html#Code_fontification_with_Gnus_and_Org-mode

That's cool.  We should publicize it more, it's extremely useful,
especially if it takes off and other MUAs support it, and such bells and
whistles are always popular with the Gnus users.  Can you explain more,
with examples for other languages (and how to add more if they are
missing)?

Thanks
Ted

Re: quoting Emacs Lisp

[Julien Danjou]

[-- Attachment #1.1: Type: text/plain, Size: 1002 bytes --]

On Thu, Jan 20 2011, Ted Zlatanov wrote:

> That's cool.  We should publicize it more, it's extremely useful,

Isn't it! :)

> especially if it takes off and other MUAs support it, and such bells and
> whistles are always popular with the Gnus users.  Can you explain more,
> with examples for other languages (and how to add more if they are
> missing)?

Well, the block is extracted by Gnus and displayed inline as an org-mode
buffer (see commit 2d2f709ab067c5688f325fc64c618988ed6fd178 for
implementation detail, it's really simple).

The font lock is then done by org-mode itself with the help of Babel[1].

Note that attaching file inline (or making them rendered inline with
`K i') is now also supported by Gnus if the file has a correct
content-type. I've added for example shell script colorization with
commit 6242139e39fe923bf469377ed1d64487c900d0eb. You just have to
copy/paste this for other language you usually find in attached file.

This should be fontified:


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1.2: arf.sh --]
[-- Type: text/x-sh, Size: 24 bytes --]

# Some comment
echo foo

[-- Attachment #1.3: Type: text/plain, Size: 106 bytes --]



[1]  http://orgmode.org/worg/org-contrib/babel/

-- 
Julien Danjou
❱ http://julien.danjou.info

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: quoting Emacs Lisp

[Julien Danjou]

> This should be fontified:

With 6d5f40566b89879bee46366b5232c65d28bec457 of course.
(it seems application/x-sh and text/x-sh both exists and I ignored that.
:-))

-- 
Julien Danjou
❱ http://julien.danjou.info

Re: Problem displaying multipart/alternative messages

[Lars Ingebrigtsen]

Michael Piotrowski <mxp@dynalabs.de> writes:

>> Debugger entered--Lisp error: (error "Selecting deleted or non-existent buffer")
>>   mm-preferred-alternative-precedence(((#<buffer " *mm*<2>">

This should now probably be fixed, although since I don't have any test
cases that trigger this bug, I'm not quite sure.

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

Re: quoting Emacs Lisp

[Lars Ingebrigtsen]

Julien Danjou <julien@danjou.info> writes:

> I was just trying my recent pushed patches.[1]
>
> [1]  http://julien.danjou.info/blog/2011.html#Code_fontification_with_Gnus_and_Org-mode

Neat.  That explains the colourful code patches I were seeing.  :-)

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

Re: Problem displaying multipart/alternative messages

[Julien Danjou]

[-- Attachment #1: Type: text/plain, Size: 461 bytes --]

On Sat, Jan 22 2011, Lars Ingebrigtsen wrote:

> Michael Piotrowski <mxp@dynalabs.de> writes:
>
>>> Debugger entered--Lisp error: (error "Selecting deleted or non-existent buffer")
>>>   mm-preferred-alternative-precedence(((#<buffer " *mm*<2>">
>
> This should now probably be fixed, although since I don't have any test
> cases that trigger this bug, I'm not quite sure.

I confirm it is fixed.

-- 
Julien Danjou
❱ http://julien.danjou.info

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: Problem displaying multipart/alternative messages

[Eric S Fraga]

Julien Danjou <julien@danjou.info> writes:

> On Sat, Jan 22 2011, Lars Ingebrigtsen wrote:
>
>> Michael Piotrowski <mxp@dynalabs.de> writes:
>>
>>>> Debugger entered--Lisp error: (error "Selecting deleted or non-existent buffer")
>>>>   mm-preferred-alternative-precedence(((#<buffer " *mm*<2>">
>>
>> This should now probably be fixed, although since I don't have any test
>> cases that trigger this bug, I'm not quite sure.
>
> I confirm it is fixed.

Seems to be fixed for me as well.  Thanks!

-- 
: Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.0.50.1 + No Gnus v0.11

Re: quoting Emacs Lisp

[Lars Ingebrigtsen]

Julien Danjou <julien@danjou.info> writes:

> This should be fontified:
>
> # Some comment
> echo foo

When I view this article, I get the following message in the minibuffer:

Indentation setup for shell type bash

Which is kinda, er, unexpected when you're reading an email.

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

Re: Problem displaying multipart/alternative messages

[Lars Ingebrigtsen]

Julien Danjou <julien@danjou.info> writes:

> #+begin_src emacs-lisp
>     (unless (with-current-buffer (mm-handle-buffer handle)
>       &))
> #+end_src

And this message flashes

OVERVIEW

in the minibuffer.

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

Re: quoting Emacs Lisp

[Julien Danjou]

[-- Attachment #1: Type: text/plain, Size: 368 bytes --]

On Mon, Jan 24 2011, Lars Ingebrigtsen wrote:

> When I view this article, I get the following message in the minibuffer:
>
> Indentation setup for shell type bash
>
> Which is kinda, er, unexpected when you're reading an email.

That's shell-script-mode being verbose. I can't see what to do about it. :)

-- 
Julien Danjou
❱ http://julien.danjou.info

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: quoting Emacs Lisp

[Ted Zlatanov]

On Mon, 24 Jan 2011 11:29:40 +0100 Julien Danjou <julien@danjou.info> wrote: 

JD> On Mon, Jan 24 2011, Lars Ingebrigtsen wrote:
>> When I view this article, I get the following message in the minibuffer:
>> 
>> Indentation setup for shell type bash
>> 
>> Which is kinda, er, unexpected when you're reading an email.

JD> That's shell-script-mode being verbose. I can't see what to do about it. :)

I can push a fix into the Emacs trunk.  Looks like just `sh-set-sell'
needs a fix for this issue; I can add a `sh-verbose' variable you can
override.  `message' is used unconditionally a LOT in that code but I
don't think it's necessary to make it all dependent on `sh-verbose'.

WDYT?

Ted

Re: quoting Emacs Lisp

[Lars Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> I can push a fix into the Emacs trunk.  Looks like just `sh-set-sell'
> needs a fix for this issue; I can add a `sh-verbose' variable you can
> override.  `message' is used unconditionally a LOT in that code but I
> don't think it's necessary to make it all dependent on `sh-verbose'.

Sounds good.  Or perhaps all the messages in sh-set-shell should just be
removed.  Do you really need to know "Indentation setup for shell type
bash" every time you open a shell script?

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

Re: quoting Emacs Lisp

[Ted Zlatanov]

On Mon, 24 Jan 2011 13:47:27 -0800 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> I can push a fix into the Emacs trunk.  Looks like just `sh-set-sell'
>> needs a fix for this issue; I can add a `sh-verbose' variable you can
>> override.  `message' is used unconditionally a LOT in that code but I
>> don't think it's necessary to make it all dependent on `sh-verbose'.

LI> Sounds good.  Or perhaps all the messages in sh-set-shell should just be
LI> removed.  Do you really need to know "Indentation setup for shell type
LI> bash" every time you open a shell script?

That message can be useful.  Properly solving the problem would require
a numerical sh-verbose and assigning levels to the important messages
(there are several that are errors and should be shown regardless).

CC to the author, Daniel Pfeiffer <occitan@esperanto.org>, to see what
he thinks.

Ted

Re: quoting Emacs Lisp

[Daniel Pfeiffer]

  la 01/24/2011 11:09 PM Ted Zlatanov skribis:
> CC to the author, Daniel Pfeiffer <occitan@esperanto.org>, to see what
> he thinks.

No opinion.  I switched to Perl scripting many years ago.

coralament / best Grötens / liebe Grüße / best regards / elkorajn salutojn
Daniel Pfeiffer

-- 
lerne / learn / apprends / lär dig / ucz się    Esperanto:
                     http://lernu.net  /  http://ikurso.net

Re: quoting Emacs Lisp

[Julien Danjou]

[-- Attachment #1: Type: text/plain, Size: 711 bytes --]

On Mon, Jan 24 2011, Ted Zlatanov wrote:

> I can push a fix into the Emacs trunk.  Looks like just `sh-set-sell'
> needs a fix for this issue; I can add a `sh-verbose' variable you can
> override.  `message' is used unconditionally a LOT in that code but I
> don't think it's necessary to make it all dependent on `sh-verbose'.

I personally do not care that much about messages {ab,}use, and I do not
know the {ab,}use rules that might exist in the Emacs development
process.

It's mostly Lars that seems to be strict with message use, so I let you
with your Emacs developer hat and Lars with its
I-really-care-about-messages hat decide. ;-)

-- 
Julien Danjou
❱ http://julien.danjou.info

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]