ietf-nntp OVER extension

[Andrew Gierth]

>>>>> "Russ" == Russ Allbery <rra at stanford.edu> writes:

 >> There was a suggestion that OVER could take a message-ID. This
 >> would need to be optional; to allow it, include the lines prefixed
 >> with $, while to forbid it exclude those lines.

 Russ> Why does this need to be optional?  It's a new command; we can
 Russ> require that this be implemented.  I'm personally slightly
 Russ> inclined to require it, but I think if it's not required, it
 Russ> should be omitted entirely.  Adding an optional command is just
 Russ> too much more complexity.

I'm somewhat against having OVER <id>. It adds relatively little
functionality as compared to HEAD, and it doesn't really make sense
from an implementation point of view either in the client or the
server.

(for one thing, the process of taking an article header and turning
it into overview-format data is normally done when the article is
stored, which is likely to be an entirely separate program, rather
than when it it read.)

 >> This memo defines two metadata items: "line:count" and
 >> "byte:count".

 Russ> I'd like to see :lines and :bytes here instead.

I agree with Russ

 >> 9.5.2 The OVER Extension

 >> This extension provides two commands, OVER and LIST
 >> OVERVIEW.FMT. The label for this extension is OVER. If the
 >> extension is implemented then both commands MUST be provided.

 Russ> I think we should drop the LIST OVERVIEW.FMT command.  As
 Russ> discussed on this list rather extensively, it doesn't provide
 Russ> any useful information, and with the introduction of metadata
 Russ> names it just confuses things.

I agree with Russ here too.

 >> 9.5.2.1.1 Responses

 >> $      First form (optional range specified):

 >> 224   Overview information follows (multi-line response)
 >> 412   No newsgroup currently selected
 >> 423   No article(s) selected

 Russ> That should be 420, and you're missing the 430 (no such article
 Russ> number) response here.

420 is "no current article", i.e. you did OVER without specifying a
range, where no current article pointer is set. 423 would be if you
specified an article number or range that didn't exist (though see
other discussion).

430 is only for when a message-id is specified.

-- 
Andrew.