- From: Doug Davis <dug@us.ibm.com>
- Date: Wed, 31 Oct 2001 16:58:51 -0500
- To: Noah_Mendelsohn@lotus.com
- Cc: henrikn@microsoft.com, hugo@w3.org, jacek@systinet.com, xml-dist-app@w3.org
I think it would be interesting to see how many SOAP processors would be able to successfully process a message where instead of: <env> <header> <myheader1 mu=1/> <header> <body> <getQuote.../> </body> </env> we sent: <env> <body> <myheader/> <getQuote.../> </body> </env> Just a guess but I think most would not - and I think it is mainly because they had a similar interpretation of the spec. When we say headers should be (or are) just like the body, why do we say that headers should be used as extensions - why not use the body? Sure we can, as long as you wanted an implicit MU attribute - but we don't tell people to do that - should we? Maybe if we did that then I would be less inclined to look at the Body as a single unit. And going one step further why not just get rid of the MU attribute at all - then we have Martin's proposal, where we have an optional section and a mandatory section - or if we want to have an ordered mixing of MU and non-MU blocks then just get rid of the HEADER and BODY elements and just have blocks. I think it is important to understand why the above example would fail (my assumption) and what we could do to fix it so that either we don't break them or we make it perfectly clear to implementors why they're all wrong. -Dug Noah_Mendelsohn@lotus.com on 10/31/2001 03:29:20 PM To: Doug Davis/Raleigh/IBM@IBMUS cc: henrikn@microsoft.com, hugo@w3.org, jacek@systinet.com, xml-dist-app@w3.org Subject: RE: Comments on issue 101 Dug Davis writes: >> The spec specifically says we're not going to >> deal with things like boxcarring so what >> does multiple independent >> body blocks mean then, if not taken a single >> unit ? I think it means the same thing as multiple header blocks, all with mustUnderstand="1", all addressed to the anonymous actor. That is: follow the rules in chapter 2. Just because you understand and process multiple entries, whether in Header or Body, does not mean your are boxcarring (it might, we can't prohibit it, but we aren't working through the many issues you would need to handle to make boxcarring work.) In general, we say: Note also that nothing in the spec says anything about the order of processing of such anonymous header blocks vs. body blocks. They're symmetric. In general, order is indeterminate unless the message uses special features: [1] "It is possible that the processing of particular SOAP block would control or determine the order of processing for other SOAP blocks. For example, one could create a SOAP header block to force processing of other SOAP header blocks in lexical order. In the absence of such a SOAP block, the order of processing is at the discretion of the SOAP node." I really think we don't want to invent special rules for all of this involving the body. It's syntactic sugar for a common case, and it is suggestive of being the main point of the message. Also: I disagree with those who say that the body can't be manipulated by intermediaries. If that were true, how could we do encryption at intermediaries? [1] http://www.w3.org/TR/2001/WD-soap12-part1-20011002/#procsoapmsgs ------------------------------------------------------------------------ Noah Mendelsohn Voice: 1-617-693-4036 Lotus Development Corp. Fax: 1-617-693-8676 One Rogers Street Cambridge, MA 02142 ------------------------------------------------------------------------
Received on Wednesday, 31 October 2001 17:00:09 UTC