Re: article editing oddity

[Shenghuo ZHU <zsh@cs.rochester.edu>]

>>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

LMI> Vladimir Volovich <vvv@vvv.vsu.ru> writes:
>> i've got a message which was in windows-1251 encoding, but
>> Content-Type header said that it was in koi8-r, so the message was
>> unreadable.
>> 
>> I pressed C-u g e, and changed "charset=koi8-r" to
>> "charset=windows-1251" in the Content-Type header.
>> 
>> Then i pressed C-c C-c
>> 
>> But the saved article still contained "charset=koi8-r" !

LMI> Uhm.  `C-c C-c' after editing the message will create brand new
LMI> Content-Type headers.  And since Gnus thought that the text was
LMI> koi8-r, Gnus put that in the new header.

How about not rebuilding headers if the article is shown by `C-u g'?

Or not reuilding headers when check in by `C-u C-c C-c'?

LMI> *sigh*

LMI> Anyway, `C-u W M c' will let you re-interpret the message using a
LMI> different charset.

[...]

I don't think this works, unless press `C-u g C-u W M c', since it
washes an article in the article buffer, where characters have been
decoded by (wrong) charset, then the result is double-decoded,
i.e. `\201's shows.

Even `C-u g C-u W M c' does not decode subject and other headers.

*sigh*

My unofficial solution is advicing gnus-summary-show-article. When
number argument comes, pick a charset from predefined charsets or
minibuffer, set gnus-newsgroup-charset and
gnus-newsgroup-ignored-charsets, then ad-do-it.

If this solution is acceptable, I can patch the
gnus-summary-show-article, and maybe a little work let gnus ignored
all charsets.

-- 
Shenghuo ZHU