[Mailman-Developers] multipart/alternatives and Archive Scrubber behaviour

[Steve Lay]

MM 2.1.5

After complaints from some list members about the text of their messages
going missing when viewed in the archives I've noticed some slightly
undesirable behaviour in the scrubber module.

The problem concerns people who regularly send multipart/alternative
messages containing text/plain and text/html parts.  The scrubbing policy
for the site is correctly implemented for these people until the user
decides to send an attachment.  At that point the message becomes a
multipart/mixed message containing two a multipart/alternative and some
other attachment, typically application/msword or some such.

The scrubber uses the "walk" method of an email message to find and scrub
those hard-to-shift HTML stains but when the archiver page is generated in
the second pass the scrubber uses the get_payload method instead.  As a
result, the scrubber never descends into the nested multipart/alternative
part and doesn't display either the text of the message or a link to the
HTML attachment (your results will vary depending on your site policy for
the archiver).

Is this a bug or is there a real reason why get_payload is used for the
second stage of the scrubber process?

Steve Lay