8-bit headers not encoded

8-bit headers not encoded

[Roman Belenov]

Hello,

In  5.8.2 8-bit headers were correctly MIME encoded. In 5.8.3 they are
sent as is and it seems that encoding is not right.

Best regards,
 Roman                          mailto:roman@nstl.nnov.ru

Re: 8-bit headers not encoded

[Florian Weimer]

Roman Belenov <roman@nstl.nnov.ru> writes:

> In  5.8.2 8-bit headers were correctly MIME encoded. In 5.8.3 they are
> sent as is and it seems that encoding is not right.

Probabile mea culpa.  Which newsgroups, which charset?  Could you give
a Message-ID of an affected article, please?

Re[2]: 8-bit headers not encoded

[Roman Belenov]

[-- Attachment #1: Type: text/plain, Size: 1554 bytes --]

Hello Florian,

FW> Roman Belenov <roman@nstl.nnov.ru> writes:

>> In  5.8.2 8-bit headers were correctly MIME encoded. In 5.8.3 they are
>> sent as is and it seems that encoding is not right.

FW> Probabile mea culpa.  Which newsgroups, which charset?  Could you give
FW> a Message-ID of an affected article, please?

Well, I just tested it internally (sent mail to myself). I'm using GNU
Emacs   20.5   under   Windows   NT;   here  are  my  charset-related
settings  (probably  most  of  them are not necessary) :

(setq   gnus-group-charset-alist '((".*" koi8-r)))
(setq   gnus-group-posting-charset-alist '((".*" koi8-r)))
(setq gnus-default-charset 'koi8-r)
(setq gnus-default-posting-charset 'koi8-r)
(setq message-default-charset 'koi8-r)
(setq default-sendmail-coding-system 'koi8-r)
(setq sendmail-coding-system 'koi8-r)
(setq-default sendmail-coding-system 'koi8-r)

(codepage-setup 1251)
(set-terminal-coding-system 'cp1251)
(setq default-buffer-file-coding-system 'cp1251-dos)
(set-selection-coding-system 'cp1251-dos)
(set-default-font
      (create-fontset-from-ascii-font
       "-*-Lucida Console-normal-r-*-*-13-*-*-*-*-*-iso8859-5"))

Also I have the following in my ~/.gnus

(put-charset-property 'cyrillic-iso8859-5 'prefered-coding-system 'koi8-r)


Attached  are received mail messages (the subject and the body contain
word  'test' in Russian). Both versions correctly encoded message body
to koi8-r, although subject was sent in cp1251 by 5.8.3 version.


Best regards,
 Roman                            mailto:roman@nstl.nnov.ru

[-- Attachment #2: 5.8.3.txt --]
[-- Type: text/plain, Size: 844 bytes --]

Received: from nstl.nstl.nnov.ru (nstl.nstl.nnov.ru [195.98.58.2]) by bsd.nstl.nnov.ru (8.9.3/8.9.3) with
ESMTP id OAA39343 for <roman@nstl.nnov.ru>; Thu, 6 Jan 2000 14:10:15 +0300 (MSK) (envelope-from
roman@nstl.nnov.ru)
Received: from LIGHTHOUSE (nc034.nstl.nnov.ru [195.98.58.34] (may be forged)) by nstl.nstl.nnov.ru
(8.9.3/8.9.3) with ESMTP id NAA11636 for <roman@nstl.nnov.ru>; Thu, 6 Jan 2000 13:41:50 +0300 (MSK/MSD)
To: roman@nstl.nnov.ru
Subject: òåñò
From: Roman Belenov <roman@nstl.nnov.ru>
Date: 06 Jan 2000 14:06:25 +0300
Message-ID: <u1z7v31e6.fsf@nstl.nnov.ru>
Lines: 3
User-Agent: Gnus/5.0803 (Gnus v5.8.3) Emacs/20.5
MIME-Version: 1.0
Content-Type: text/plain; charset=koi8-r
Content-Transfer-Encoding: 8bit
Status:   

ÔÅÓÔ
-- 
                                                        With regards, Roman.

[-- Attachment #3: 5.8.2.txt --]
[-- Type: text/plain, Size: 861 bytes --]

Received: from nstl.nstl.nnov.ru (nstl.nstl.nnov.ru [195.98.58.2]) by bsd.nstl.nnov.ru (8.9.3/8.9.3) with
ESMTP id OAA39300 for <roman@nstl.nnov.ru>; Thu, 6 Jan 2000 14:03:04 +0300 (MSK) (envelope-from
roman@nstl.nnov.ru)
Received: from LIGHTHOUSE (nc034.nstl.nnov.ru [195.98.58.34] (may be forged)) by nstl.nstl.nnov.ru
(8.9.3/8.9.3) with ESMTP id NAA11617 for <roman@nstl.nnov.ru>; Thu, 6 Jan 2000 13:34:39 +0300 (MSK/MSD)
To: roman@nstl.nnov.ru
Subject: =?koi8-r?b?1MXT1A==?=
From: Roman Belenov <roman@nstl.nnov.ru>
Date: 06 Jan 2000 13:59:13 +0300
Message-ID: <u9023ea9q.fsf@nstl.nnov.ru>
Lines: 3
User-Agent: Gnus/5.0802 (Gnus v5.8.2) Emacs/20.5
MIME-Version: 1.0
Content-Type: text/plain; charset=koi8-r
Content-Transfer-Encoding: 8bit
Status:   

ÔÅÓÔ
-- 
                                                        With regards, Roman.

Re: Re[2]: 8-bit headers not encoded

[Florian Weimer]

Roman Belenov <roman@nstl.nnov.ru> writes:

> (setq   gnus-group-posting-charset-alist '((".*" koi8-r)))

Could try the following value for gnus-group-posting-charset-alist,
please?

(setq gnus-group-posting-charset-alist 
  '((message-this-is-mail nil nil)
    (message-this-is-news koi8-r koi8-r)))

This should mimic the behavior of the old setting.  (Perhaps this
variable should be given an new name?)

Re[4]: 8-bit headers not encoded

[Roman Belenov]

Hello Florian,

FW> Could try the following value for gnus-group-posting-charset-alist,
FW> please?

FW> (setq gnus-group-posting-charset-alist 
FW>   '((message-this-is-mail nil nil)
FW>     (message-this-is-news koi8-r koi8-r)))

FW> This should mimic the behavior of the old setting.  (Perhaps this
FW> variable should be given an new name?)

Yes,  with this setting GNUS 5.8.3 works fine; it works fine also when
I  don't  change  this  variable  at  all in my .emacs, so I was wrong
trying to play with it. I've set it long ago time when I tried to make
GNUS  use  koi8-r  instead of iso8859-5 when posting. Now all relevant
information  is  in the manual under "Charsets" topic (it wasn't there
several months ago).

I don't think that this variable should be renamed, since it can break
working configs. Maybe some words should be included in documentation,
like "don't try to use this variable to change the charset of outgoing
messages, use 'preferred-coding-system property instead".


Best regards,
 Roman                            mailto:roman@nstl.nnov.ru

Re: Re[2]: 8-bit headers not encoded

[Vladimir Volovich]

"FW" == Florian Weimer writes:

 FW> Could try the following value for
 FW> gnus-group-posting-charset-alist, please?

 FW> (setq gnus-group-posting-charset-alist '((message-this-is-mail
 FW> nil nil) (message-this-is-news koi8-r koi8-r)))

according to doc, the third element should be a list of charsets (or t
or nil); so `koi8-r' is incorrect there.

 FW> This should mimic the behavior of the old setting.  (Perhaps this
 FW> variable should be given an new name?)

i tried the following setting:

(setq gnus-group-posting-charset-alist 
      '((message-this-is-mail koi8-r t)
	(message-this-is-news koi8-r t)))

Then, 8-bit cyrillic text in subject was sent as 8-bit, but NOT in
koi8-r; it was in iso-8859-5 encoding. what should i do to make gnus
use koi8-r for 8-bit text in headers?

note that this behavior changed (was broken) in gnus 5.8.3; gnus 5.8.2
worked correct (i.e. 8-bit text in headers was sent using koi8-r).

	Best regards, -- Vladimir.

Re: Re[2]: 8-bit headers not encoded

[Kai Großjohann]

Vladimir Volovich <vvv@vvv.vsu.ru> writes:

> "FW" == Florian Weimer writes:
> 
>  FW> (setq gnus-group-posting-charset-alist '((message-this-is-mail
>  FW> nil nil) (message-this-is-news koi8-r koi8-r)))
> 
> according to doc, the third element should be a list of charsets (or t
> or nil); so `koi8-r' is incorrect there.

Well, `koi8-r' is not a list.  I don't want to appear pedantic, but
I'm afraid Gnus might be...

kai
-- 
A large number of young women don't trust men with beards.  (BFBS Radio)

Re: Re[2]: 8-bit headers not encoded

[Florian Weimer]

Vladimir Volovich <vvv@vvv.vsu.ru> writes:

>  FW> (setq gnus-group-posting-charset-alist '((message-this-is-mail
>  FW> nil nil) (message-this-is-news koi8-r koi8-r)))
> 
> according to doc, the third element should be a list of charsets (or t
> or nil); so `koi8-r' is incorrect there.

Yes, I should have written `(koi8-r)' instead of `koi8-r'.

> Then, 8-bit cyrillic text in subject was sent as 8-bit, but NOT in
> koi8-r; it was in iso-8859-5 encoding. what should i do to make gnus
> use koi8-r for 8-bit text in headers?

Okay, I'll try to fix it as soon as I find some time (unfortunately,
that's a big problem at the moment...).  In the meantime, could you
resist to set this variable at all?  The default value seems to work
correctly (it uses quoted-printable in the header).

BTW, do you specify `iso-8859-5' somewhere as default?  I wonder where
that charset is coming from.

Re: Re[2]: 8-bit headers not encoded

[Vladimir Volovich]

"FW" == Florian Weimer writes:

 >> Then, 8-bit cyrillic text in subject was sent as 8-bit, but NOT in
 >> koi8-r; it was in iso-8859-5 encoding. what should i do to make
 >> gnus use koi8-r for 8-bit text in headers?

 FW> Okay, I'll try to fix it as soon as I find some time
 FW> (unfortunately, that's a big problem at the moment...).  In the
 FW> meantime, could you resist to set this variable at all?

well, when posting to some newsgroups, i need to use 8-bit koi8-r
encoded text in headers (e.g. subject) because otherwise cyrillic text
will be messed up when gatewaying them e.g. to FidoNet.

so, i do need this feature :-) (meanwhile, i could use gnus 5.8.2)

 FW> The default value seems to work correctly (it uses
 FW> quoted-printable in the header).

to be pedantic, gnus uses for iso-8859-5 not quoted-printable, but `b'
type encoded-word (which is closer to base64 than to quoted-printable).

 FW> BTW, do you specify `iso-8859-5' somewhere as default?

Not much, but i wish i had a possibility to specify preferable
cyrillic charsets for message body on a per-group basis. currently,
gnus insists on using a charset which is defined in a line like

(put-charset-property 'cyrillic-iso8859-5 'preferred-coding-system 'koi8-r)

for all cyrillic postings. I'd like to be able to specify a preferred
cyrillic charset on a per-group basis (e.g. like
gnus-group-posting-charset-alist). this would be useful not only for
cyrillic, but also for other languages which have several popular
charsets in use. for russian, we have at least five charsets: koi8-r,
windows-1251, cp866, mac-cyrillic, iso-8859-5 (there are even more of
them: e.g. cp855; the total number of variants is more than 20
different 8-bit cyrillic charsets!).

then one may wish to use by default, say, a windows-1251 charset for
some newsgroups, and koi8-r otherwise, etc...

for european latin-based languages, there are also several charsets in
use (e.g., ISO latin-N charsets and windows codepages)...

 FW> I wonder where that charset is coming from.

it is coming from ISO. :-) It was once a russian state standard
GOST19768-74, but later, koi8-r (and in some areas, cp866, and then
cp1251) was adopted as a de-facto standard.

	Best regards, -- Vladimir.