ietf-nntp HDR

[Russ Allbery]

Ade Lovett <ade at lovett.com> writes:

> Been there, done that.  I think that's why we rewrote Lines headers for
> the past 10 years or so, in various disguises.

> You are falling into the classic trap of working around something that
> needs to be addressed elsewhere, namely in the posting of articles,
> where it should simply state that the server MUST ignore the
> client-supplied Lines: header (on POST) and generate its own (accurate)
> header.

And you're making a bad engineering decision by inlining data that's
already available in an out-of-band form with guaranteed accuracy, as well
as a second bad engineering decision of putting derived data into a header
where it can get out of sync.  cf Content-Length instead of From as a
delimiter in Unix mailbox format.

Now that we've both stated our strongly-held opinions on this matter, can
I point out that this is completely out of scope for the NNTP working
group, that in practice Lines is *not* standardized, that your proposal
was already shot down on USEFOR, and that regardless we cannot assume that
your proposal has been implemented for the purposes of this standard?

There was a clear consensus on USEFOR against trying to standardize the
Lines header, and a clear consensus in favor of declaring it obsolete.  In
this case, given the people who were involved in that particular
consensus, I'll be quite surprised if it's changed.

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