- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Wed, 30 Aug 2006 14:07:32 +0100
- To: Ashok Malhotra <ashok.malhotra@oracle.com>
- Cc: "public-ws-policy@w3.org" <public-ws-policy@w3.org>
I'm a bit confused by this desiderata. There seem to be two casees mentioned: 1) The need for a determinate order for signing/verification. In this case the order doesn't *mean* anything. That is, we can have equivalent policies that differ merely by the order of assertions in an <all>. To ensure that we are using the authenticated policy, obviously we need the signed one, but, in a sense, we should be free to substitute any equivalent one. 2) Order with semantic import. I don't know what this is except to look at the example. On Aug 29, 2006, at 3:25 PM, Ashok Malhotra wrote: [snip] > For example, consider an assertion that adds something to a message. > Perhaps a timestamp. We may want to say that the timestamp is > added before a log record is written. [snip] But I have trouble wrapping my head around an *assertion* that does anything but state something :) Side-effecting assertions seem like real trouble. Consider Prolog vs. SQL. It could (must) mess with distribution and other properties of operators. Cheers, Bijan.
Received on Wednesday, 30 August 2006 13:08:27 UTC