Re: nntp-extensions Re: ietf-nntp NNTP SEARCH extension internet-draft available

[Chris Newman <[e-mail]>]

On Thu, 31 Oct 1996, Brian Hernacki wrote:
> I agree that when replicating functionality into NNTP that exists in
> IMAP we should attempt to keep as similar as practical. I think we did
> this reasonably well with the proposed search extension.

Here's my input on your search extension:

1) Don't muck with the way IMAP did the charsets for searching.  Choosing
iso-8859-1 as the default is completely wrong for the russian newsgroups
where koi8-r is the default.  Why not just keep the [CHARSET charset_name]
grammar from IMAP with the default CHARSET of us-ascii?  It won't
kill the Europeans to make the "CHARSET iso-8859-1" explicit, and it will
make the Russians and others much happier.  Using MIME-2 strings just
makes things unnecessarily complex -- who would want to search with
multiple character sets at the same time?  That's rare enough that the
extra code (and associated speed loss) necessary to make it work is a
lose.  It took us months to settle on the CHARSET mechanism in IMAP's
search command -- please don't make us go through the debate again.  It's
implementable and meets international demand.

2) Why not include the SMALLER and LARGER search terms?  Quite useful for
detecting MIME-grams and things not to download over a slow link.

3) Why not keep "UID", so you can search over a range of
articles within a single newsgroup?  It'd be somewhat silly for a search
across multiple newsgroups, but that's not a big deal.

4) Why not enclose the <group_pat> in parentheses rather that double
quotes?  It's a set of things -- not a single thing.

5) You're missing the "BODY" search term in the grammar.

6) Why put the string "HEADER" at the beginning?  It confuses the syntax
with the "HEADER" search token, and is technically wrong since a BODY or
TEXT search can follow.

7) It wouldn't hurt to copy the language from the IMAP spec which allows
SEARCH to ignore non-text body parts.

8) it wouldn't hurt to try to make your ABNF more precise.

> However, what about other mechanisms which do not exist in IMAP but
> exist or are need in NNTP? NEWGROUPS and (the proposed) LIST
> SUBSCRIPTIONS, and LIST MOTD for example?

Those need to be added, but not necessarily to the IMAP protocol. The ACAP
protocol is much better suited to managing large lists of stuff, and would
allow transparent distribution of newsgroups and mailboxes across multiple
IMAP servers.  A single ACAP mailbox list dataset could trivially include
the functionality of NEWGROUPS and most of the NNTP LIST commands with a
much more powerful access paradigm.