Keith Moore <moore(_at_)cs(_dot_)utk(_dot_)edu> writes:
Not NNTP. The NNTP RFC was published before there were any real
implementations, which is one of the reasons why it has so many flaws
that later had to be corrected in practice.
thanks for correcting my mis-impression.
though it's certainly possible for protocols that are widely deployed
prior to being published to have large numbers of flaws also ...
Agreed. However, NNTP is for me an excellent example of why it's good to
write at least *one* implementation before publishing the standard
(incidentally, the author of the NNTP standard agrees wholeheartedly with
me on that and would do it differently in retrospect). There were ideas
in there that sounded good on paper but in practice simply didn't work,
like SLAVE.
--
Russ Allbery (rra(_at_)stanford(_dot_)edu)
<http://www.eyrie.org/~eagle/>