[NNTP] Re: MODE READER

[Ken Murchison]

Andrew - Supernews wrote:

>>>>>>"Mark" == Mark Crispin <mrc at CAC.Washington.EDU> writes:
> 
> 
>  Mark> The cases then degenerate into:
> 
>  Mark> 1) MODE READER isn't required
> 
> This is, by far, the most common case.
> 
>  Mark> 2) MODE READER is required, but you must authenticate before
>  Mark> you can do any reading.  The extension for MODE READER is not
>  Mark> listed until after authentication.
> 
> That's backwards. Where MODE READER is required, it has to come
> _before_ authentication (or TLS). I should have spotted that error in
> your previous post. (Implementations where MODE READER is an issue
> almost certainly won't be advertising auth or TLS in the LIST
> EXTENSIONS output until after the MODE READER command, and even if
> they do, the MODE READER has to come first because authentication
> is likely to be handled differently for transit clients.)

Given INN's use of MODE READER, I would expect this to be the case. 
Assuming that this is the expected use, I think we can address these 
issues with 2 capabilities (these are actually arguments to the NNTPv2 
extension label since we're talking about base functionality):

1. MODE_READER - (as already proposed by Clive) specifies that MODE 
READER is required before using any reader commands

2. MODE_FEEDER (or FEED_ONLY or NO_READING, etc) - specifies that the 
server doesn't permit *any* reader commands

The absence of both of these options means that the server permits 
reader commands and MODE READER isn't required.  The two options MUST 
NOT be advertised at the same time.

The only case which isn't addressed is a reader-only server, which 
really only effects IHAVE and I think returning a 503 solves this.  I'm 
guessing most reading clients wouldn't bother with IHAVE anyways.

-- 
Kenneth Murchison     Oceana Matrix Ltd.
Software Engineer     21 Princeton Place
716-662-8973 x26      Orchard Park, NY 14127
--PGP Public Key--    http://www.oceana.com/~ken/ksm.pgp