From: Russ Allbery (rra@stanford.edu)
Date: Mon Jun 12 2000 - 16:56:06 CDT
Maurizio Codogno <mau@beatles.cselt.it> writes:
> I agree that message IDs SHOULD NOT be longer than 250 octects
> (rather than characters).
This is sufficient for me. SHOULD NOT is sufficient (at least to me) to
use as justification for rejecting messages with longer message IDs. (And
yes, octets should be used rather than characters, agreed.)
> I may even agree that M-ID MUST NOT be longer than 250 octects,
> unless someone shows that there are good reasons to raise that limit.
I have a slight preference for going to MUST NOT, and I think the
interoperability problems, at least with exceeding 497 octets, warrant
that.
> And I have no problem in adding to the draft a note which says that
> a message ID longer than 497 octects will be refused by NNTP.
That's also a reasonable way of stating the problem.
> But since there is written nowhere than NNTP is the only way to exchange
> netnews articles, I strongly object to use that fact as a reason to put
> *that* limit.
Hm. Given how widespread use of NNTP is, I think that we do need to give
interoperability problems with NNTP a lot of weight. That's not the same
as saying that news will always be transferred via NNTP, just that we do
recognize that we need to interoperate with it.
-- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>