Still problems with non-ascii subjects

RT Developers
#1

Hello!

I installed rt-3.0.3rc1.

We are still having troubles with non-ascii Subjects.

  1. If Subject contains non-ascii characters and is plain 8bit, it
    becomes ‘???’ after RT processing (both at WebUI and in e-mail RT resends
    to subscribers). According to RT logs, encoding is determined correctly by
    Encode::Guess. The affected e-mail can be found at
    http://mitya.pp.ru/rt-bad-email1.dat

  2. Imagine that Subject contains 2 words (with non-ascii characters) delimeted
    with space in the middle that are properly MIME-encoded
    (for instance, =?koi8-r?B?1MXT1CDUxdPU?= ).
    In WebUI, this Subject is being displayed correctly. But in the e-mail
    RT sends, it looks like =?KOI8-R?B?1MXT1A==?= =?KOI8-R?B?1MXT1A==?=

So each word is encoded separatly. In that case space between the words
should also be encoded (like ‘_’ or smth else, I don’t remember).

As a result, MUA displays these two words WITHOUT space between them.

Is it possible to address these issues?

Thanks!!

#2

Dmitry Sivachenko wrote:

  1. If Subject contains non-ascii characters and is plain 8bit, it
    becomes ‘???’ after RT processing (both at WebUI and in e-mail RT resends
    to subscribers). According to RT logs, encoding is determined correctly by
    Encode::Guess. The affected e-mail can be found at
    http://mitya.pp.ru/rt-bad-email1.dat

Are headers allowed to contain 8bit data? My recollection, and
reading of RFC 822 and RFC 2047, gives me the impression that
they’re not.

Note, also, that SpamAssassin also flags that subject as “illegal”.

#3

Dmitry Sivachenko wrote:

  1. If Subject contains non-ascii characters and is plain 8bit, it
    becomes ‘???’ after RT processing (both at WebUI and in e-mail RT resends
    to subscribers). According to RT logs, encoding is determined correctly by
    Encode::Guess. The affected e-mail can be found at
    http://mitya.pp.ru/rt-bad-email1.dat

Are headers allowed to contain 8bit data? My recollection, and
reading of RFC 822 and RFC 2047, gives me the impression that
they’re not.

Note, also, that SpamAssassin also flags that subject as “illegal”.

Standards are much simpler than real life ;-((
There are many MUAs that send 8bit data in headers.
Think of standard UNIX mail(1) for instance.

#4

Dmitry Sivachenko wrote:

  1. If Subject contains non-ascii characters and is plain 8bit, it
    becomes ‘???’ after RT processing (both at WebUI and in e-mail RT resends
    to subscribers). According to RT logs, encoding is determined correctly by
    Encode::Guess. The affected e-mail can be found at
    http://mitya.pp.ru/rt-bad-email1.dat

Are headers allowed to contain 8bit data? My recollection, and
reading of RFC 822 and RFC 2047, gives me the impression that
they’re not.

Headers preferably should not contain 8bit data (un-encoded), for fear of
causing certain not-mentioned MTAs much grief, but MTAs should be able to
accept 8bit headers.

( Be conservative in what you send, liberal in what you accept ).

                         Bruce Campbell                            RIPE
               Systems/Network Engineer                             NCC
             www.ripe.net - PGP562C8B1B             Operations/Security