[NNTP] Consensus?
Steve Walker
nntp at nntpserver.com
Fri Aug 12 09:23:01 PDT 2005
Clive D.W. Feather wrote:
> I don't accept that just telling people to use 2^64-1 is a solution.
> Existing clients *ARE NOT* going to be updated "just like that", and any
> proper solution needs to work reasonably well with such clients.
Could you explain how it would cause any existing client or
server to break that wasn't already going to break at the same
point in time? My point, as well as several others, is that a 32
bit client and a 64 bit server will quit working at the exact
same time that the 32 bit server overflows. If we advocate using
64 bit numbers then at least some clients will work when the time
comes. The current wording causes all clients to fail.
> This is not "a minor change" and I don't believe we can just sneak it into
> the document without wider review.
It wasn't a minor change to add the 32 bit wording either, yet it
was added. If you really want to leave the issue open, the new
32 bit limit need to be removed from the proposed RFC977
replacement. By asking people to address this issue in a future
RFC you are effectively telling software developers to use no
specified limit (RFC977), then use a 32 bit limit (V2), then use
xyz limit (V3). It should be either fixed correctly or not
changed. You seem to be ignoring the fact that there are popular
64 bit clients in use today. The new wording causes existing
software to become non-compliant.
Steve.
More information about the ietf-nntp
mailing list