bug in v5.10.2: gnus-group-charset-alist is ignored for some messages

bug in v5.10.2: gnus-group-charset-alist is ignored for some messages

[Vladimir Volovich]

[-- Attachment #1: Type: text/plain, Size: 800 bytes --]

Hi,

i'm using Gnus v5.10.2 from CVS

i have set the default charset for reading messages in
gnus-group-charset-alist to koi8-r instead of iso-8859-1:

(eval-after-load "gnus-sum"
  '(setq gnus-group-charset-alist
     (append (delete '(".*" iso-8859-1) gnus-group-charset-alist)
	'((".*" koi8-r)))))

this works most of the time but fails for some articles.

i've attached the message for which gnus shows iso-8859-1 in the body,
instead of koi8-r. i think that gnus somehow tries to detect the
charset of the message body (and the body contains some character
unencodable in koi8-r which makes it think that this is iso-8859-1),
which it should not do if there's no charset specified in
content-type, and it should use the values specified in
gnus-group-charset-alist (and maybe others).

Best,
v.

[-- Attachment #2: message.gz --]
[-- Type: application/x-gzip, Size: 1521 bytes --]

Re: bug in v5.10.2: gnus-group-charset-alist is ignored for some

[Vasily Korytov]

On Sat, 03 May 2003 19:27:12 +0400, Vladimir Volovich wrote:

> (eval-after-load "gnus-sum"
>   '(setq gnus-group-charset-alist
>      (append (delete '(".*" iso-8859-1) gnus-group-charset-alist)
> 	'((".*" koi8-r)))))

FYI, there's no ".*" value in gnus-group-charset-alist as of Oort (IMHO,
you should'n touch it at all in your case). There is
gnus-default-charset, and it's much more easier to tweak. =))

> i think that gnus somehow tries to detect the charset of the message
> body

Gnus, not Emacs? Are you sure?

(Anyway, I'm unable to reproduce this problem on XEmacs/non-MULE -- you
can guess, why. =)

> (and the body contains some character unencodable in koi8-r which
> makes it think that this is iso-8859-1)

The message is correct KOI-8, BTW.

-- 
       I accept RFC3156 and RFC2440-compatible encrypted mail.
PGP key fingerprint: 123A 7CCE 6E26 6233 0D87 E01A A0F8 3524 FCD8 1841

Re: bug in v5.10.2: gnus-group-charset-alist is ignored for some

[Vladimir Volovich]

"VK" == Vasily Korytov writes:

 VK> On Sat, 03 May 2003 19:27:12 +0400, Vladimir Volovich wrote:
 >> (eval-after-load "gnus-sum" '(setq gnus-group-charset-alist
 >> (append (delete '(".*" iso-8859-1) gnus-group-charset-alist)
 >> '((".*" koi8-r)))))

 VK> FYI, there's no ".*" value in gnus-group-charset-alist as of Oort
 VK> (IMHO, you should'n touch it at all in your case). There is
 VK> gnus-default-charset, and it's much more easier to tweak. =))

 >> i think that gnus somehow tries to detect the charset of the
 >> message body

 VK> Gnus, not Emacs? Are you sure?

 VK> (Anyway, I'm unable to reproduce this problem on XEmacs/non-MULE
 VK> -- you can guess, why. =)

 >> (and the body contains some character unencodable in koi8-r which
 >> makes it think that this is iso-8859-1)

 VK> The message is correct KOI-8, BTW.

there was indeed a bug in gnus - ant it was fixed by Jesper Harder:

2003-05-03  Jesper Harder  <harder@ifa.au.dk>
 
        * gnus.el: Don't test for `mm-guess-mime-charset'.
 
        * mm-util.el (mm-guess-mime-charset): Remove.  Not used any more.
 
        * gnus.el (gnus-default-charset): Set default value to
        `undecided'.
 
        * gnus-art.el (article-decode-charset): Don't supply 4th arg to
        mm-decode-body.
 
        * mm-bodies.el (mm-decode-coding-region-safely): Remove.
        (mm-decode-body): Don't use mm-decode-coding-region-safely.

Best,
v.

Re: bug in v5.10.2: gnus-group-charset-alist is ignored for some

[Vladimir Volovich]

"VK" == Vasily Korytov writes:

 >> (and the body contains some character unencodable in koi8-r which
 >> makes it think that this is iso-8859-1)

 VK> The message is correct KOI-8, BTW.

when i view that message in emacs as koi8-r, it is unable to encode
one character - and shows it like \277 (it's apparently the numero
sign); since it is shown as \277, it looks like a bug in Emacs'
definition of koi8-r, since this character is also present in the
underlying iso-8859-5 coding system.

could someone please report that bug?

Best,
v.