ietf-nntp Proposed new HDR text

[Russ Allbery]

Charles Lindsey <chl at clw.cs.man.ac.uk> writes:
> Russ Allbery <rra at stanford.edu> writes:

>> ...  In particular, I have, for the time being, dodged the entire issue
>> of metadata and specified a command that always and exclusively returns
>> the actual values of the headers of the articles.

> Fair enough, though I think we ought to consider that possibility in due
> course.

My preference would be for an advocate of including metadata to write up
the changes to the posted HDR text to allow for that (I think they should
be fairly simple, just a minor alteration in the syntax and an additional
paragraph or two in the description) so that we can discuss them.

> There are essentially two possibilities we are likely to encounter. The
> implementation supports HDR on ALL headers, or it supports just those in
> the Overview.

Unfortunately, I think that it doesn't simplify quite that well.  It looks
like the situation is more like:

    overview <= HDR headers <= all headers

and in particular there are sites that don't allow HDR on all headers (for
performance reasons) but only on those headers that they index, but they
don't return all indexed headers on overview requests to cut down on the
bandwidth usage for overview.  So things like HDR Content-Type will work,
but HDR X-This-Is-Silly won't.

The list of headers HDR can be used with is potentially quite long, so I'm
not sure LIST EXTENSIONS is the right way to document them.  I'm also not
clear on why just trying the HDR command and seeing if it fails isn't a
reasonable way of handling this; in other words, why would a client need
to retrieve the list?

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