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

[Jeffrey M. Vinocur]

On Tue, 1 Apr 2003, Russ Allbery wrote:

> Jeffrey M Vinocur <jeff at litech.org> writes:
> >> Clive D W Feather <clive at on-the-train.demon.co.uk> writes:
> 
> >>> (1) Is NEWGROUPS supposed to be consistent with LIST ACTIVE.TIMES? Is
> >>> this a SHOULD or a MUST sort of thing?
> 
> > I definitely don't want it to be MUST.  I'm not sure it should even by
> > SHOULD.  I mean, do we really want the server to be checking up on
> > whether the admin has modified one of the data files (assuming an
> > INN-like implementation) and throwing a fit if the two lists aren't
> > consistent?  That doesn't seem necessary to me.
> 
> Well, with an INN implementation it's the same file.  

Heh.  Oops.


> LIST ACTIVE.TIMES is optional, so an implementation that didn't want to
> bother could just not implement it....

True.  But, consider the following scenario...

A server has all of the creation dates available for the past decade.  
Everyone is happy if NEWGROUPS returns all the groups since any date, no 
matter how old.  But to avoid wasting bandwidth, the admin would like LIST 
ACTIVE.TIMES to only return groups created in the past year.

I don't see why the network protocol specification would want to prohibit
this behavior.


-- 
Jeffrey M. Vinocur
jeff at litech.org