[Mailman-Developers] more 2.1b3++ MIME funnies.
Barry A. Warsaw
barry@python.org
Mon, 14 Oct 2002 17:25:46 -0400
>>>>> "CVR" == Chuq Von Rospach <chuqui@plaidworks.com> writes:
CVR> I'm finding that if someone sends a M/A message with a plain
CVR> and an html part, it's getting processed, and I'm getting a
CVR> message going out with two t/p parts, one the original, and
CVR> one empty. Since the empty one is second, it gets displayed.
CVR> So something is whacked out. I'm not sure what.
Odd. I just tried this w/cvs. No blacklist, whitelist as you
described, then I sent a m/a with a text/plain first and text/html
second. In this configuration, I got a message with just the
text/plain part, as I'd expect. Switch the two and I get the lynx'd
text/html part as expected.
Remember that one content filtering rule is that after filter/pass but
before html conversion, multipart/alternative parts are replaced with
their first non-empty subpart.
CVR> What I'd like to be able to do (but can't) is allow M/A,
CVR> t/plain and t/html, and use the lynx to defang HTML going
CVR> into the text (but not MIME) digests. That seems the most
CVR> reasonable set of compromises to allow styled
CVR> mail. Unfortunately, mailman can't be set up that way...
Nope, because messages are processed for the two digests in the same
fell swoop. I don't think it's ever come up before that you might
want different processing for the different digests, but that /does/
make sense. ("Oh, TODO, please come here for a moment")
CVR> so my backup plan was to allow text/plain, convert html to
CVR> text if the text part didn't exist, and try to add features
CVR> later. But I don't seem to be quite right yet. For what I'm
CVR> trying, I *think* I should be whitelisting M/A and
CVR> Text/Plain, but not text/HTMl. that screws people who only
CVR> send text/html without a plain part, but they probably
CVR> shouldn't do that anyway...
That's the setting that I think most people were talking about when
content filtering was added.
I think things are working properly, although the feature may not be
ideally useful yet. I knew this one would be tough to make flexible
enough.
-Barry