- From: <noah_mendelsohn@us.ibm.com>
- Date: Sat, 24 Aug 2002 18:31:16 -0400
- To: "Henrik Frystyk Nielsen" <henrikn@microsoft.com>
- Cc: Mike.Champion@SoftwareAG-USA.com, xml-dist-app@w3.org
Well, I'm not so sure regarding the sender. We don't really say anything about the steps leading to the preparation of the envelope. We just say what is SHOULD/MUST/SHOULD NOT/MUST NOT contain. I would rather not get into a two step description along the lines of: you might erroneously put in a PI but then you should check to see whether you did. I think we should just say: senders MUST NOT send PIs, Intermediaries detect and remove PIs. Receivers SHOULD fault when receiving PIs. Note, however, that intermediaries MAY but need not detect (and fault) when a PI is received in a message to be relayed; this dispensation is provided primarily to facilitate the implementation of high performance intermediaries in which such checking may be impractical. Such intermediaries MAY relay PIs received in the inbound message (but MUST NOT introduce additional or altered PIs.) ------------------------------------------------------------------ Noah Mendelsohn Voice: 1-617-693-4036 IBM Corporation Fax: 1-617-693-8676 One Rogers Street Cambridge, MA 02142 ------------------------------------------------------------------ "Henrik Frystyk Nielsen" <henrikn@microsoft.com> 08/23/02 08:13 PM To: <noah_mendelsohn@us.ibm.com>, <Mike.Champion@SoftwareAG-USA.com> cc: <xml-dist-app@w3.org> Subject: RE: Problem with resolution of Issue 221 I agree with this but would also go further in stating that this seems to apply to any SOAP sender, regardless of whether it is the initial sender or an intermediary sender: for performance reasons, it would be really bad for a sender to first go through the message and check for PIs before sending. If we want to say anything for PIs then I think it should be SHOULD. FWIW, I would be happy not to say anything. Henrik >My strong feeling is that intermediaries should not be >required to do PI >checking in situations where performance makes such detection >a problem. I >agree with Gudge that requiring it for one purpose but not >another misses >the point. So, if the resolution to 221 seems inconsistent >when viewed >from that perspective, then I think we need to get the WG to >clarify. My >recollection was that our intention was that detection and >rejection in a >receiver was to be on a best effort basis, but I could be >wrong. Thanks.
Received on Saturday, 24 August 2002 18:32:09 UTC