The bug was only happening on some messages, and kept falling behind other work. I've stolen it back and taken a look, and it was actually a deliberate change in our parser to handle some types of broken email, though I think the change was done incorrectly.
Basically, its broken if the message/rfc822 message has a content-disposition of attachment, but works otherwise. The example message it was designed to fix attached an actual .eml file as message/rfc822 but base64 encoded it. base64 (or quoted-printable) encoding of a message/* is illegal per RFC 2045 section 6, so the correct solution was to consider the message/rfc822 not a multipart container if it was encoded, not if it was of disposition attachment. In theory, we could base64 decode the part, but that would be a much bigger change to the parser. Fix forthcoming. Brandon On Thu, Feb 23, 2012 at 7:43 AM, Tony Hansen <[email protected]> wrote: > Totally off topic: I wish I knew how to get them to fix that. I get bit by > it regularly. > > On topic: because gmail messes up in this area doesn't mean that we should > also mess things up in the same way. > > I am a firm believer in the use of message/rfc822 attachments. > > Tony Hansen > > > On 2/23/2012 9:51 AM, Jan Kundrát wrote: >> >> On 02/20/12 11:52, Tony Finch wrote: >>> >>> message/rfc822 is also good. IMO every message that is forwarded or >>> top-post replied should use message/rfc822 attachments rather than doing >>> some fucked up redaction of the original message. >> >> GMail still breaks [1] on such messages -- the web interface shows the >> message source and the IMAP server returns invalid BODYSTRUCTURE which >> doesn't allow accessing the attached message. >> >> Cheers, >> Jan >> >> [1] http://permalink.gmane.org/gmane.mail.imap.general/2637 > > > _______________________________________________ > imap5 mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/imap5 _______________________________________________ imap5 mailing list [email protected] https://www.ietf.org/mailman/listinfo/imap5
