ietf-nntp Virtual hosts in NNTP servers

[Jeffrey M. Vinocur]

On Wed, 26 Feb 2003, Russ Allbery wrote:

> Joao Prado Maia <jpm at papercut.org> writes:
> 
> > They will obviously ask the reason for them needing to enter an
> > 'anonymous' type user and mark 'My server requires authentication' in
> > something that doesn't really require authentication - that's the
> > definition of anonymous access, after all.
> 
> It's a hack, but it's a hack that's also widely used by other protocols
> with this same issue.

On the other hand, the fact that it requires every user configuring 
another option is a tad ugly.


> I don't think anyone's saying that a HOST command
> wouldn't be cleaner, just that it turns out not to be strictly necessary.

I agree it's not necessary, but for different reasons :-)

I'm still not convinced that virtualhosting like this is really that 
useful.  I mean, if an anonymous connection from a particular IP is going 
to be able to access certain groups with one hostname, and certain other 
groups on the other hostname, I really don't see the benefit to separating 
them -- if anything, it's a bit reminiscent of security-by-obscurity.  The 
only case where this wouldn't work is if newsgroup names aren't unique 
across virtualhosts (e.g. if you have two *different* local.support 
groups, or something), but I can't imagine any server has support for that 
either.  In other words, the *content* of the servers is never 
conflicting, so the union of access permissions would work just fine.

On the other hand, if the consensus is that the view I describe above is 
silly (and I'm certainly not passionate about this!), I'd say it's worth 
making an extension for this.


> I have no objections to someone writing up a HOST extension that behaves
> as you specify.  

Since I'm going to have to resubmit the STARTTLS draft after IETF, it 
seems, I can probably have this done by then as well.

I'll separate it out from the STARTTLS stuff into a separate document, I 
think.  The more the merrier, right?

-- 
Jeffrey M. Vinocur
jeff at litech.org