bug#45925: 27.1; *Summary* buffer vs. raw utf-8 headers

["積丹尼 Dan Jacobson" <jidanni@jidanni.org>]

>>>>> "LI" == Lars Ingebrigtsen <larsi@gnus.org> writes:
LI> 積丹尼 Dan Jacobson <jidanni@jidanni.org> writes:

>> Try this simple experiment:
>> $ echo Subject: 一二三|procmail
>> $ echo Subject: 一二三|iconv -t big5|procmail

LI> I don't have procmail installed, so I'm not sure what these do -- are
LI> you sending a mail (to yourself?) here?  Do you have a recipe to
LI> reproduce this problem without the use of procmail?

$ echo Subject: 一二三 > ~/Maildir/new/Z
$ file ~/Maildir/new/Z
~/Maildir/new/Z: UTF-8 Unicode text


>> $ emacs -f gnus
>> 
>> In the *Article* buffer, both look like
>> Subject: 一二三
>> In the *Summary* buffer so does the big5 version.
>> Alas, the utf-8 version looks like
>> c\x80\xd3....
>> 
>> (Yes, these are illegal raw headers. But Gnus is supposed to be
>> accommodating. And it does... but oddly not for the majority (UTF-8) case.)

LI> [...]

>> (Might be related to bug#45724.)

LI> Is this still with nnml?  If so, could you find the resulting lines in
LI> the .overview files in the nnml directory and post them here?  (Perhaps
LI> after gzipping them to avoid Emacs helpfully re-encoding the lines.)

Yes, nnml.

The headers get appended raw to .overview.

Thus .overview contains a mix of ASCII, big5, and UTF-8, all in the same file.

$ echo Subject: 一二三|iconv -t big5 > ~/Maildir/new/B5
$ echo Subject: 一二三 > ~/Maildir/new/UT
$ emacs -f gnus
$ tail -n 2 Mail/mail/misc/.overview|qprint -e
37397   =A4@=A4G=A4T    (nobody)                <87a6t4gnpx.5.fsf@totally-fudged-out-mess=
age-id>         0       0       Xref: jidanni5 mail.misc:37397=09
37398   =E4=B8=80=E4=BA=8C=E4=B8=89     (nobody)                <878s8ognpx.5.fsf@totally-=
fudged-out-message-id>          0       0       Xref: jidanni5 mail.misc:37398=09

Anyway: *Summary* oddly can only deal with raw big5, not raw UTF-8.
However *Article* can deal with both.