[NNTP] Capability modifiers

[Russ Allbery]

Please, anyone with any sort of opinion on this, speak up.  We need to
make a decision on it as I believe it's the only issue holding us up from
getting out a final version of this draft and going into a working group
last call on our documents.

Clive D W Feather <clive at demon.net> writes:

> If we remove modifiers entirely, we still want to say *something* about
> extensibility of the CAPABILITIES syntax, and we still need something to
> replace the -MODE-READER modifier in particular.

As noted by others, I think the VERSION line offers sufficient
extensibility in the general case, and for this in particular it seems
like one could safely introduce modifiers later by stating that servers
MUST NOT send lines beginning with characters not allowed in capabilities
and clients MUST ignore any such lines if present.  That would allow
modifiers to be added as a separate experimental provision and adopted if
they prove useful in practice.

I would not replace -MODE-READER with anything; I would require readers to
issue a MODE READER command if they want to use READER facilities, those
facilities are not advertised, and MODE-READER is.

I think most people know my opinion on this, but to restate it just for
the record:

> * Is the modifier *concept* worth retaining as a hook for syntax
> extension?  If not, how do we address that?

I don't think it's sufficiently useful to be worth pursuing at this time.

> * Is the -- modifier useful as a documentation technique?

I find -- to be the most useful of the lot, as it's a way of clearly
indicating "posting will never be permitted" that's better than the 201
greeting code.  But that isn't sufficiently interesting to have the
modifier be worth it for me.

> * Is the -label modifier useful for interrelated capabilities?

> * Are the -480 and -483 modifiers useful for indicating what is and
> isn't available with and without security and privacy?

I like -480 and -483 better than -label, but none of them seem
particularly useful to me.  I think the client has sufficient information
to make good decisions for all of the common cases, and I'm not sure
modifiers really help the uncommon cases that much.

> * If we remove modifiers, what do we do about -MODE-READER?

Akin to authentication and encryption, I would have the server simply not
advertise capabilities that are not currently available but that will
become available after MODE READER.

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