- From: Henrik Frystyk Nielsen <frystyk@microsoft.com>
- Date: Thu, 3 May 2001 08:36:50 -0700
- To: "'Doug Davis'" <dug@us.ibm.com>
- Cc: "'Martin Gudgin'" <marting@develop.com>, "XML Protocol Comments" <xml-dist-app@w3.org>
>>Making a high level hint about the content tightly bound to the >>specific instance of a specific message is fundamentally a bad idea. >>Henrik > >I believe you've also stated that routing done based on >the SOAPAction is also a bad idea and yet people are doing it >today because the spec doesn't say they can't or even >shouldn't. Do you think we should add those restrictions? This is a rather different point. We can define the semantics of features but never prevent them from being used in other ways. What we are talking about here is what the semantics of the field is. >Having the SOAPAction tightly bound to the specific content >of the message is a perfect valid choice (it doesn't violate >the spec) and IMHO the benefits gained are enormous [1]. Exactly - you pick the right word - "choice". SOAP doesn't dictate what the value should be just as well as HTTP doesn't dictate what the content type should be and SMTP doesn't dictate who the recipient of a mail should be. Regardless of whether the value is tightly bound to a specific element or not, one has to know the value just as well as one has to know the encoding of the SOAP message, the endpoint for which it is intended, the parameters that one are to send etc. Either way it doesn't matter because the whole idea is that whoever wants to use the hint doesn't have to walk through the body so whether it is tightly bound to a specific element in the envelope or not is unimportant to the user of the hint - the purpose is exactly for the user *not* to care. Henrik
Received on Thursday, 3 May 2001 11:37:35 UTC