[NNTP] Consensus?

[Charles Lindsey]

In <42FB52EC.5070002 at oceana.com> Ken Murchison <ken at oceana.com> writes:

>Then the client chokes and can't read the group.  They are no worse off 
>then if the server supports your BIGNUM strawman (which I like BTW) and 
>the client doesn't.  The bottom line is that if we recommend 2^64 in 
>this spec, it gives client authors at *least* two years to upgrade the 
>code.  If they don't upgrade (to 2^64 OR BIGNUM) by the time article 
>numbers on servers exceed 2^32, I'd consider the clients horribly broken.

But it is totally naive to expect that anything approaching the majority
of clients will be upgraded within two years, even if every known
implementation produces an upgraded version within that timescale, which
is itself most unlikely.

In USEFOR, there are umpteen schemes which we have had to discard because
the client base cannot be assumed to have changed. Even schemes where we
had hoped that the base of servers might be upgraded within a couple of
years are now being removed from the draft (expectations that folded
Newsgroups headers could be used within two years, or so, for example).

And in Email, they go to enormous lengths to ensure that all clients MUST
accept strange things which have not been seen in real emails for many
years, and most of which, I suspect, have never been used in anger at all.

That seems to be the IETF way of doing things. The only way you can get
away with something which will not interoperate with existing software (or
which cannot be worked around with existing software) is to define an
Experimental Protocol, and I don't think that is the way forward for the
present problem.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl at clerew.man.ac.uk      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5