191885 - charset and hence content of subject changed upon reply/followup (different charset in subject and body)

Status: VERIFIED DUPLICATE of bug 115096

(MailNews Core :: MIME, defect)

Description

[Boris 'pi' Piwinger]

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b) Gecko/20030117
(also seen in Windows version)

If I reply to a mail/posting with the following subject, it is changed to
iso-8859-1:
Subject: =?ISO-8859-5?Q?=C7=E2=DE_=DF=E0=DE=D1=DB=D5=DC=D0=3F?=

Of course, the new subject does not make any sense.

In this test case the body has been in ASCII. If the body is utf-8 the Subject
just becomes:
Subject: Re: =?UTF-8?B?57+9?=

Everything is OK, if the body has iso-8859-5.

If the body is iso-8859-1, then we get the same result as for ASCII.

The error already happens when composing. This bug is evil. If you correct the
subject by hand, Mozilla readily sends everything, so we could do it right.

And BTW: This breaks GNKSA which requires that the subject is not changed.

pi

Comment 1

[John Moreno]

Yes, this was recently discussed on USEFOR -- the Subject shouldn't 
have it's charset changed unless the user changes content of the header.

Except for the addition of "Re: " if necessary and the possible deletion of 
"(was: .+", the Subject that comes in should byte for byte the Subject that 
goes out, unless the user makes a change (at that time your free to 
change the encoding to fit what the user is using).

(I'd also recommend raw UTF8 instead of the 2047 encoding shown, but 
that's a different matter).

Comment 2

[Boris 'pi' Piwinger]

>Yes, this was recently discussed on USEFOR -- the Subject shouldn't 
>have it's charset changed unless the user changes content of the header.

Well, this is not the point of my report. If it would be a correct recoding
(like changing ISO-8859-5 to UTF-8) that would be OK. This change without
recoding does change the content.

>Except for the addition of "Re: " if necessary and the possible deletion of 
>"(was: .+", the Subject that comes in should byte for byte the Subject that 
>goes out, unless the user makes a change (at that time your free to 
>change the encoding to fit what the user is using).

Which would also mean to preserve real bad subjects (unnecessary binary
encoding, MIME where it is not needed, bad folding, mixed charsets etc.).
Anyway, if you want it, please use a new bug (if there is none so far).

>(I'd also recommend raw UTF8 instead of the 2047 encoding shown, but 
>that's a different matter).

I don't know any RfC which would allow that. Maybe if USEFOR comes to some
result, which I personally would not expect before 2137, probably later. But
again: Not this bug.

pi

Comment 3

[Igor Velkov]

looks like duplicate of 115096

Comment 4

[Boris 'pi' Piwinger]

You are right.

BTW: If you write "bug 115096" instead of only the number, it becomes clickable.

pi

*** This bug has been marked as a duplicate of 115096 ***

Comment 5

[Stephen Donner [:stephend] Not actively reading bugmail]

Verified dup