RE: Soap Message Canonicalization (SM-C14N)

>Suggestion:  "SOAP requires that 'header blocks be processed in
>a manner fully conformant with the specification for that block' [1].
>Such specifications MUST indicate (a) the situations, if any in which
>an intermediary is to 're-insert' a header block into a relayed
>SOAP message and (b) any constraints on the position within the
>relayed header at which the re-insertion is to occur." 
>Editorial wordsmithing is required, but I think this is the
>right approach.

I tend to see this more as a guideline than a hard, testable
requirement. If we want to provide such guidelines then it seems to be
applicable for SOAP nodes in general and not particularly for
intermediaries. From a SOAP point of view I don't think there is any
special treatment of re-inserted header fields in this respect. Would
this be the sort of material suited for the primer?

>>> * There are no restrictions on where a SOAP intermediary can
>>> insert additional header blocks. That is, an intermediary can
>>> insert blocks between any other blocks in a SOAP header.
>
>Again, I think the proper positioning of a header block should be 
>per the specification for that block.  As always, if multiple 
>modules are used together, the specs must explain how to use
>the modules together - in this case, how to position all of 
>the required blocks.  I think this is true both at the original
>sender of a message, and at an intermediary inserting new
>headers.

I think the intent of this text is to cover a different scenario in
which header blocks do not know about each other. If should be possible
to deploy orthogonal header blocks within the same message that do not
have any knowledge about each other.

>>> Hope this makes sense!
>
>Yes, absolutely.  Do my suggested changes also make sense?

Yes, thanks!

Henrik

Received on Wednesday, 20 February 2002 15:37:12 UTC