changing gnus-group-posting-charset-alist

changing gnus-group-posting-charset-alist

[Katsumi Yamaoka]

[-- Attachment #1: Type: text/plain, Size: 2006 bytes --]

Hi,

I'm wondering for years about the default value for the
gnus-group-posting-charset-alist variable.

| gnus-group-posting-charset-alist's value is
| (("^\\(no\\|fr\\)..." iso-8859-1 (iso-8859-1))
|  ("^\\(fido7\\|relcom\\)..." koi8-r (koi8-r))
|  (message-this-is-mail nil nil)
|  (message-this-is-news nil t))
|
| Alist of regexps and permitted unencoded charsets for posting.
| Each element of the alist has the form (TEST HEADER BODY-LIST), where
| TEST is either a regular expression matching the newsgroup header or a
| variable to query,
| HEADER is the charset which may be left unencoded in the header (nil
| means encode all charsets),
| BODY-LIST is a list of charsets which may be encoded using 8bit
| content-transfer encoding in the body, or one of the special values
| nil (always encode using quoted-printable) or t (always use 8bit).
|
| Note that any value other than nil for HEADER infringes some RFCs, so
| use this option with care.

Is it still necessary not to encode text by qp or something to
post news articles to no, fr, fido7 and relcom newsgroups?  I
don't use Norwegian, French or Russian, so it isn't my business,
though.

As for other newsgroups, Gnus forces the 8bit encoding especially
to news articles' body by default.  Isn't it a vestige of the
time when Gnus had not been multiligualized?  It is hard for me
to imagine there's a difference between mail and news.  What I'd
like to say is: let's change the default value to:

(message-this-is-news nil nil)

Almost Japanese text which I write can be encoded by iso-2022-jp
and 7bit.  So, those text will be posted as 7bit articles.
However, I sometimes followup to articles containing characters
which cannot be encoded by 7bit.  In those cases, text will be
encoded by shift_jis, utf-8 and so forth and fed to the server
without being encoded with qp or base64.  A newsreader posting
such articles may be only Gnus nowadays.  Therefore, I make it a
rule to recommend Japanese users to modify the value as follows:


[-- Attachment #2: Type: application/emacs-lisp, Size: 211 bytes --]

[-- Attachment #3: Type: text/plain, Size: 7 bytes --]


WDYT?

Re: changing gnus-group-posting-charset-alist

[Jesper Harder]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> What I'd like to say is: let's change the default value to:
>
> (message-this-is-news nil nil)
>
> WDYT?

The posting guidelines for the dk.* hierarchy discourage CTE=QP.  So
at least for that case it's not a good change.

-- 
Jesper Harder                                <http://purl.org/harder/>

Re: changing gnus-group-posting-charset-alist

[Katsumi Yamaoka]

>>>>> In <m3is9cdiwz.fsf@defun.localdomain>
>>>>>	Jesper Harder <harder@ifa.au.dk> wrote:

> Katsumi Yamaoka <yamaoka@jpl.org> writes:

>> What I'd like to say is: let's change the default value to:

>> (message-this-is-news nil nil)

>> WDYT?

> The posting guidelines for the dk.* hierarchy discourage CTE=QP.  So
> at least for that case it's not a good change.

I see.  Thanks.  I'll reverse the approach.  The main groups to
which Japanese articles are posted are fj and japan, and gmane.
All gmane groups allow the mail convention.  So, I think I can
add the following elements as the default value.

 ("Regexp for fj and japan" nil nil)
 ("Regexp for Gmane groups" nil nil)

Is there an evil thing?

Re: changing gnus-group-posting-charset-alist

[Katsumi Yamaoka]

>>>>> In <yotlbrf3v9t9.fsf@jpl.org> Katsumi Yamaoka wrote:

> All gmane groups allow the mail convention.

I meant it is nonsense that Gnus forces the 8bit encoding (it is
neither qp nor base64) to articles to be posted to Gmane groups
by default.  It is because I think those groups should be
treated as mail groups.  If there's no counter evidence, I'll
add a rule to gnus-group-posting-charset-alist so that Gnus may
impose no restraints on the encoding of articles to be posted to
Gmane groups.  And I'll do the same to Japanese newsgroups.

Re: changing gnus-group-posting-charset-alist

[Jesper Harder]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> | gnus-group-posting-charset-alist's value is
> | (("^\\(no\\|fr\\)..." iso-8859-1 (iso-8859-1))
> |  ("^\\(fido7\\|relcom\\)..." koi8-r (koi8-r))
> |  (message-this-is-mail nil nil)
> |  (message-this-is-news nil t))
>
> Is it still necessary not to encode text by qp or something to
> post news articles to no, fr, fido7 and relcom newsgroups?

I think the reason for the special entries isn't CTE=QP as such, but
that those hierarchies don't want rfc2047-encoded headers.

> As for other newsgroups, Gnus forces the 8bit encoding especially
> to news articles' body by default.  Isn't it a vestige of the
> time when Gnus had not been multiligualized?  It is hard for me
> to imagine there's a difference between mail and news.

The major difference is that NNTP is 8bit-clean whereas SMTP isn't.
Or more precisely: The NNTP draft standard states that NNTP is
8bit-clean (which has been true for a long time).  The SMTP standard
says you shouldn't use 8bit in email except when the server supports
the ESMTP extension 8BITMIME.

In practice SMTP probably _is_ 8bit-clean today -- but we wouln't want
a heavily armed IETF squad team knocking at the door because we're
violating RFC 2821.  Which is why we use CTE=QP.

> Almost Japanese text which I write can be encoded by iso-2022-jp
> and 7bit.  So, those text will be posted as 7bit articles.
> However, I sometimes followup to articles containing characters
> which cannot be encoded by 7bit.  In those cases, text will be
> encoded by shift_jis, utf-8 and so forth and fed to the server
> without being encoded with qp or base64.  A newsreader posting
> such articles may be only Gnus nowadays.

It's interesting that your perspective is the opposite of mine: I see
QP mainly as an annoying concession to maybe a few old SMTP servers
which still aren't 8bit-clean.  If possible I'd prefer 8bit
everywhere, because you can use the text directly without decoding it
first, for example you can grep your mailbox directly.

I'm not saying that you're wrong, but that this preference is probably
locale-dependent.

-- 
Jesper Harder                                <http://purl.org/harder/>

Re: changing gnus-group-posting-charset-alist

[Katsumi Yamaoka]

The scales dropped from my eyes.  That's a Japanese expression
meaning that misunderstandings were removed from my mind (in
this case, scale is a skin of a big fish).

>>>>> In <m3sm8b69xw.fsf@defun.localdomain> Jesper Harder wrote:

> Katsumi Yamaoka <yamaoka@jpl.org> writes:

>>| gnus-group-posting-charset-alist's value is
>>| (("^\\(no\\|fr\\)..." iso-8859-1 (iso-8859-1))
>>|  ("^\\(fido7\\|relcom\\)..." koi8-r (koi8-r))
>>|  (message-this-is-mail nil nil)
>>|  (message-this-is-news nil t))
>>
>> Is it still necessary not to encode text by qp or something to
>> post news articles to no, fr, fido7 and relcom newsgroups?

> I think the reason for the special entries isn't CTE=QP as such, but
> that those hierarchies don't want rfc2047-encoded headers.

I see.  Using such charsets is an agreement in those groups,
isn't it?

>> As for other newsgroups, Gnus forces the 8bit encoding especially
>> to news articles' body by default.  Isn't it a vestige of the
>> time when Gnus had not been multiligualized?  It is hard for me
>> to imagine there's a difference between mail and news.

> The major difference is that NNTP is 8bit-clean whereas SMTP isn't.
> Or more precisely: The NNTP draft standard states that NNTP is
> 8bit-clean (which has been true for a long time).  The SMTP standard
> says you shouldn't use 8bit in email except when the server supports
> the ESMTP extension 8BITMIME.

Indeed.  I noticed we can send anything by news, specifying a
charset explicitly or implicitly.  The value `t' will mean Gnus
is not restricted to 7bit rather than forcing 8bit into Gnus.
Now I think the value is very just.

> In practice SMTP probably _is_ 8bit-clean today -- but we wouln't want
> a heavily armed IETF squad team knocking at the door because we're
> violating RFC 2821.  Which is why we use CTE=QP.

I see.  Supposing there're still MTAs that don't handle 8bit,
what do you think about posting articles to Gmane groups?  I got
to feel that it does not matter. ;-)

>> Almost Japanese text which I write can be encoded by iso-2022-jp
>> and 7bit.  So, those text will be posted as 7bit articles.
>> However, I sometimes followup to articles containing characters
>> which cannot be encoded by 7bit.  In those cases, text will be
>> encoded by shift_jis, utf-8 and so forth and fed to the server
>> without being encoded with qp or base64.  A newsreader posting
>> such articles may be only Gnus nowadays.

> It's interesting that your perspective is the opposite of mine: I see
> QP mainly as an annoying concession to maybe a few old SMTP servers
> which still aren't 8bit-clean.  If possible I'd prefer 8bit
> everywhere, because you can use the text directly without decoding it
> first, for example you can grep your mailbox directly.

That's right.  Anyway, I appreciate having made me enlightened.

> I'm not saying that you're wrong, but that this preference is probably
> locale-dependent.

And it is perhaps personal-ism-dependent, now I think.