[NNTP] Article number wording

[Ade Lovett]

Matthias Andree wrote:
> Ade Lovett <ade at lovett.com> writes:
>>Article numbers are most
>>certainly *an* index, but they are not *the* index.
> 
> In SQL language, consider a group a table (all of the same schema =
> columns), one row per article.
> 
> The article number is THE de-facto index, ALTHOUGH it is only unique per
> (server or site, group) pair. Most NNTP clients access articles by
> (group, article number) tuple, NOT by Message-ID. Regardless of the
> Message-ID being globally unique for two years or something like that.

Please read what I wrote, article numbers are *an* index, not *the* index.

> Message-ID is used to reference articles, not to access them in the
> first place.

Huh?  ARTICLE|HEAD|BODY|STAT <msgid> must be a figment of my imagination
then.

> Paying attention to interoperability is important, and as you say, not
> standardizing is hurting implementors, give a clear and strong
> guideline, with "MUST" language, 

We are hurting implementors considerably more by not releasing the
document for publication with this eleventh-hour discussion.  That is
not to say that this discussion is in any way invalid, but having read
the entire thread, it seems very unlikely that any kind of consensus is
going to be reached quickly.  Taking one point from the above, I would
have to completely disagree on the use of MUST.  At most, it whould be a
SHOULD.

I again appeal to the WG chair to publish the document as-is, and we can
continue these discussions in the meantime, along with any other nits
that will result in the far bigger audience reading the document as a
result of it being published as a Proposed Standard.

-aDe