[Mailman-Developers] Re: Dates

David Champion dgc@uchicago.edu
Tue, 1 May 2001 23:53:49 -0500

On 2001.05.01, in <15087.30907.613868.332848@anthem.wooz.org>,
	"Barry A. Warsaw" <barry@digicool.com> wrote:
> Phew!  A little thing like dates gets us geeks in an uproar.
> Whodathunk? :)

Well, we're touching on a matter of fidelity of a record, which is what
gets me concerned about it.  I guess that's a kind of geekiness, but
maybe a different kind. :)

> First, Mailman will not rewrite Date: headers for messages exploded to
> the list.  I'm a firm believer in the least-munging rule.  If the
> message gets sent out dated in 1982, then that's what the recipients
> will see.  So that one's off the table.

Right, and, as a comment on our comments thus far, I think it's safe to
say that we all agree on it.

> Second, /all/ we're talking about is placing the article in the
> archives.  Ideally, the original Date: header would be preserved in
> the archive.  But on the other hand, it really doesn't help people
> find information when a message is placed 8 years into the future.
> Mailman's current approach is less than ideal, and I think the right
> thing to do is for the archiver to put some sanity checks on the date
> for collation purposes.  What Mailman and the receiving MTA can do is

Here we brush by the unspoken third approach to the problem.  Rather
than altering the date in the copy of the message fed to the archiver,
Mailman can feed the "correct" date out-of-band.  The archiver can use
this altered date, as you say, for collation, but retain the original
Date: header for the archive itself.

This is certainly more trouble to implement, and arguably infeasible
since it could involve modification of the (external?) archiver, but I
think it best meets all interests.

I'd also like to offer that although, for example, PGP/MIME (and other,
nonstandard, PGP message encodings for e-mail) does not sign the Date:
field along with the message itself, this is not an unreasonable thing
for some alternative or future signature format to do.  That's just
another reason that altering the archival contents themselves strikes
me as a Bad Thing, even though altering the way the message is collated
or indexed is fine.

 -D.	dgc@uchicago.edu	NSIT	University of Chicago