From: Brian McCauley (B.A.McCauley@bham.ac.uk)
Date: Sun Oct 17 1999 - 12:43:11 CDT
Russ Allbery <rra@stanford.edu> writes:
> Brian McCauley <B.A.McCauley@bham.ac.uk> writes:
>
> > But where is this written down? Draft-ietf-usefor-article-02 section
> > 7.1 implies this but it does not come out and say it (because it's
> > wrongly assumed that it should go without saying).
>
> The general consensus of the working group seems to be that submissions
> for moderated groups should be given a distinguished MIME type and
> encapsulated in a MIME e-mail message. I'm with the general consensus on
> this.
Actually so am I. At least I had independantly worked out that this
would be the "right way" if a change to the existng mechanism was
ever to be discussed. However, I'd searched the mailing list archive
quite thoroughly and concluded that the matter had not been discussed.
Was the matter discussed in some other forum?
> Encapsulating the message sounds like more work, but I really think that's
> just during the transition.
There will be a need for very careful management of the transition.
There are many ad-hoc procmail and/or perl filters out there for
moderation that would break.
I do not think it would be reasonable to hope that all moderator's
scripts could be updated before any news injection agents using the
new format were deployed.
The work-round as I see it would be to have the mail servers for
moderators.isc.org domain automatically de-encapsulate for any
newsgroup not flagged as supporting the new format. It is quite
possible that this would place an unacceptable load on the MX servers
for the moderators.isc.org domain.
To avoid lazyness causing this proceess to drag on too long I would
propose that the de-encapsulation servers append a nag line to every
text/plain article they process containing the URI of an IETF document
explaining what's going on. Yes I'm aware that this may trigger false
positives on moderation bots that check signature length but this is
still better than the same bots rejecting *all* new-style submissions
as being binaries. I'm also aware that anything that mucks about with
messaege content is likely to be unpopular with IETF but I think in
this case it's the lesser of several evils, it's only transitional and
would not be part of any standards track RFC.
It is debatable how long moderation agents should be expected to
continue to support the old format and how long the public
de-encapsualtion servers should be expected to run.