ietf-nntp HDR replacement text

[Russ Allbery]

Andrew Gierth <andrew at erlenstar.demon.co.uk> writes:
>>>>>> "Russ" == Russ Allbery <rra at stanford.edu> writes:

>  Russ> I don't think we've yet reached any conclusions about this.
>  Russ> Could other people please weigh in?  When all the articles in a
>  Russ> specified range for HDR or OVER are not present, should the
>  Russ> server return an empty list, or an error code?

> it's potentially awkward to return an error code; looking at a few
> existing implementations, returning an empty response seems to be the
> usual behaviour. (i.e. the logic is "send response; then loop through
> specified range sending data when available")

>  Russ> Specifying an individual number and specifying a range should
>  Russ> result in the same behavior.

> hrm. not sure.

> A case could possibly be made for allowing the server to return either
> the empty list or the error code.

Hm.  The more I think about it, the more that seems reasonable to me.  It
really doesn't make much difference from the client perspective, and it's
a lot easier to return an empty list in some server implementations.

Right now, I'm leaning towards either empty list or server's choice of
either, for both the case where a single number was specified but doesn't
exist and the case where nothing in a range was present on the server.

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