To:
ietf-provreg@cafax.se
From:
budi@alliance.globalnetlink.com
Date:
Wed, 3 Jan 2001 20:08:21 +0700
In-reply-to:
<4.3.1.2.20010102170144.0472c470@mail.forman.com>
Sender:
owner-ietf-provreg@cafax.se
Subject:
Re: Requirements for object existence queries
Dan Cohen wrote: > In section 3.9 [4] of the requirements draft, it says: > > A query to determine if an object exists in the registry MUST return only a > positive or negative response so that server software > that responds to this query can be optimized for speed. Pardon my ignorance (just joined a few days ago and still in ROM mode), but where can I get the complete draft? > In think that MUST should change to MAY. If a registry can implement a > highly optimized query that returns more information (maybe the > registrar who "owns" the object), so much the better. That information > would probably be quite helpful to registrars. Can't we put this under a postive response? Perhaps with a code (eg. ala FTP) that says more information is available. Then, it is up to the client to get this additional information. Perhaps it just wants to know whether such object exists or not. (Since I have not read the requirements draft, I don't know how the state diagram of the protocol looks like.) > And while we are on the topic of optimizing for speed, have people > considered that an XML based protocol might not be the ideal choice for this? > XML based protocols are almost universally more verbose than non-XML > alternatives, and while the CPU overhead for parsing the XML may be > negligible, when you consider added CPU time for encryption, plus extra > network bandwidth, you wind up which a much less performance > friendly protocol. I, too, am not too keen on XML-based protocols :-) Regards -- budi -- .ID admin -- Homepage: <http://budi.insan.co.id> my presentation materials, papers, scrapbook, ... and more What's your "web.id"? Register your web.id @ http://www.idnic.net.id