ietf-nntp Summary/analysis of LIST OVERVIEW.FMT responses

[Russ Allbery]

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

> But it appears that it is used for its intended purpose by some people.

Yes.

> The only real issue seems to be what to do if the database contents
> change.

> Proposal:

> (1) LIST OVERVIEW.FMT is optional, even if OVER is supported.

> (2) OVER output MAY, but SHOULD NOT, contain headers not listed by LIST
> OVERVIEW.FMT (that provides one path for servers to upgrade the
> database; modify the overview.fmt file only after they know the database
> is consistent once again).

> (3) If the command returns "Header:full", then absence of that header
> from OVER output means that the article does not contain that header,
> guaranteed.

This sounds good to me, actually.  I think this works with most existing
practice except with INN, and even most INN configurations.

> (4) If it returns "Header:full ?", then the absence of that header from
> OVER output does not guarantee its absence from the article (another way
> of indicating possible inconsistency).

I'd rather not do this; if we're going to keep the same command name, it
should really not change syntax.  I also don't think it's necessary, given
(1), (2), and (3).

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