I don't disagree, however, it isn't
as if there isn't an effective mitigation strategy.
The specification already covers this,
starting on line 820 (cd03):
"In order to properly secure messages,
the body and all relevant headers need to be included in the signature."
Clearly, this would be a "relevant
header", and the specification for foo:SecureRM should have a similar
caveat that strongly recommends that
the header be signed with the body.
My principal concern with a wsrm:mustUnderstand
is that it changes the parsing/processing model of the
messages from what is currently the
case for every SOAP stack. Significantly.
Adding a SOAP header that carries the
desired semantic is something that is built into every SOAP
stack. Every one of them. Today. No
changes necessary. Adding a "handler" for the mU QName
could be as trivial as adding an entry
to a configuration file of SOAP header QNames that are
"understood" by the endpoint.
Oh, the complexity. Not!
Cheers,
Christopher Ferris
STSM, Software Group Standards Strategy
email:
blog: http://www.ibm.com/developerworks/blogs/dw_blog.jspa?blog=440
phone: +1 508 377 9295
Richard Salz/Cambridge/IBM
04/25/2006 08:21 AM
To
"Gilbert Pilz" <>
cc
Christopher B Ferris/Waltham/IBM@IBMUS,
"wsrx" <>
Subject
RE: [ws-rx] issue 115: clarifying questionLink
I think Gil's got a very important point.
/r$
--
SOA Appliances
Application Integration Middleware
