W3C home > Mailing lists > Public > www-xkms@w3.org > October 2004

Non-deterministic compound requests...

From: Stephen Farrell <stephen.farrell@cs.tcd.ie>
Date: Thu, 28 Oct 2004 14:51:58 +0100
Message-ID: <4180F97E.8060506@cs.tcd.ie>
To: "Www-Xkms (E-mail)" <www-xkms@w3.org>


Hi All,

The current spec, para 77 [1] says:

"An XKMS service MAY support processing of Compound Requests. A compound
request permits multiple XKMS requests to be made at the same time. A
compound request consists of an outer request and one or more inner
requests. There is no ordering implicit in the inner requests. The
semantics of making a set of requests as a compound request are exactly
the same as if each individual request in the set had been made
separately and simultaneously."

I don't totally understand that last bit. Here's a case in point:

A compound request contains (in this order) a register request for
a key and then a locate request for something else in that binding
(say the keyname).

Assuming that all the binding information (key, keyname etc.) is
new, does the locate response indicate that the key is known or
not?

I can imagine implementations that either parallelise or sequence
(in a few possibly reasonable ways) the processing of such a
request and that the end result might be that the locate response
content is essentially unpredictable, which I dislike.

So, am I right in the above and which way would your implementation
handle that request?

Secondly, I'd imagine that there may be other nasty cases lurking
around here - can anyone think of any really bad ones?

I guess we may raise an issue on this, depending on the answers to
the above.

Stephen.

[1] 
http://www.w3.org/2001/XKMS/Drafts/XKMS-PR-DRAFT/PR-DRAFT-xkms-part-1.html#XKMS_2_0_Section_2_8
Received on Thursday, 28 October 2004 13:49:14 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 October 2009 08:39:23 GMT