ietf-nntp draft-ietf-nntpext-base-17

[Jeffrey M. Vinocur]

On Mon, 31 Mar 2003, Russ Allbery wrote:

> Clive D W Feather <clive at demon.net> writes:
> > Charles Lindsey said:
> >
> >> No, I am saying what I would _like_ the newsgroups file to contain in
> >> the future. Hence the word "SHOULD", though I could live with "should".
> 
> > Hmm. Other opinions?
> 
> I think that "SHOULD" is ruling out widespread existing practice, which I
> don't really see a justification for in this case, and "should" gets into
> the sort of halfway handwaving that I think we should avoid.  Either it's
> a protocol issue or it isn't.  If we're trying to provide useful
> documentation for people writing NNTP software, which is more what I'm
> worried about than attempts to change the world, I think we should be
> clear that LIST NEWGROUPS isn't going to return the canonical list of
> groups and they can't rely on it having any fixed relationship to the
> active file.

I agree entirely.  And it's not clear what benefit there would be from 
having the server enforce a subset relation instead of just letting the 
client do it.

(I actually have a problem with the lack of precision available from the 
2119 keywords, but c'est la vie at this point.)


-- 
Jeffrey M. Vinocur
jeff at litech.org