ietf-nntp GROUP Responses (was: Commetns on draft-15.pdf)

[Russ Allbery]

Charles Lindsey <chl at clw.cs.man.ac.uk> writes:

> AIUI, if you POST (or IHAVE) an article to comp.foo, INN stores it as if
> it had been posted to comp.bar. Or does this only apply to IHAVE, with
> POST rejecting with a 440 response?

It only applies to IHAVE.  POST is rejected.

> But the interesting question is what happens if you try to read articles
> from comp.foo, using GROUP + ARTICLE commands. You say "GROUP comp.foo",
> and what response do you get?

You get a response indicating that comp.foo is a valid group containing no
articles.  You do not see the messages for comp.bar.

Like I said, in INN it does nothing at all useful for readers.  It would
be good to flesh out aliasing so that it was actually more useful, but
it's not already been done at least in INN.

> I don't think anyone should be defining aliasing in detail at this
> point, but that is no reason not to state what responses should be
> regarded as legitimate in such seemingly bizarre situations.

I'm not sure we should introduce aliasing as a concept into our draft.
It's not already there, it's not widespread existing practice, and it's
going to be hard to explain the current half-broken state.

> All I am saying is that if you return a 211 with apparently working
> count and water marks, then it would also be better to report the group
> that the server had actually made current, whether or not that was the
> one you had asked for.

I'm sympathetic to the idea, and maybe for the use of future more useful
extensions it may be a good idea to note that the group name returned by
GROUP can be different than the group name requested.  That doesn't seem
to me to cause any harm, but I don't have a strong opinion on it either
way.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>