[NNTP] CAPABILITIES problem!
ade at lovett.com
Tue Aug 2 22:33:14 PDT 2005
Peter Robinson wrote:
> Bumping into the old limit before CAPABILITIES is even out of the door
> is a good indication that this would have been a future annoyance for
> extensions. I had assumed that the 12 character command name limit came
> from RFC977, but since that isn't the case I see no reason at all to set
> a limit.
Likewise. AIUI, the impact of having no limit is server-side, and thus
if it DOES come up as an implementation issue, is likely to be far more
solvable than a client-side change.
Indeed, I'm toying with the idea of an extension to allow for retrieval
of an article by an MD5 hash of the message ID in all places where
<messageid> is currently used. Given that such an implementation would
be of completely fixed length, it would seem to deal nicely with the
edge case of ultra-long message IDs potentially messing things up.
My vote is to remove the limit on the capabilities keywords entirely. I
could possibly be persuaded for a 64 (or even 128 character) limit, but
the current limit is far too small.
More information about the ietf-nntp