- From: Martin Gudgin <mgudgin@microsoft.com>
- Date: Sat, 6 Nov 2004 01:01:29 -0800
- To: "Glen Daniels" <gdaniels@sonicsoftware.com>, "Greg Truty" <gtruty@us.ibm.com>, <public-ws-addressing@w3.org>
> -----Original Message----- > From: public-ws-addressing-request@w3.org > [mailto:public-ws-addressing-request@w3.org] On Behalf Of Glen Daniels > Sent: 04 November 2004 20:28 > To: Greg Truty; public-ws-addressing@w3.org > Subject: RE: Issue 011 > > > > Hi Greg: > > You make a fine point below, and indeed it may be quite reasonable to > decide that <wsa:To> shouldn't be an entire EPR due to the potential > costs of copying the whole thing. Of course, those costs also exist > when serializing a <ReplyTo> or <FaultTo> EPR, which will be > very common > to do for many MEPs, or even a <From>. But ReplyTo/FaultTo/From are *fundamentally* different to To. To use the postal analogy, To is the address visible to the postman, while From/ReplyTo/FaultTo are the address(es) you put at the top of a letter. > > Perhaps this should open some questions about levels of serialization > for EPRs - for instance, if I know (through previous interaction or > out-of-band agreement) the complete (and very complex) policy for an > endpoint I'll be replying to, is it necessary for them to > serialize the > entire policy in the <ReplyTo> EPR? In a very common case (request-reply) the recipient of the message certainly knows some of the metadata related to the reply. Gudge
Received on Saturday, 6 November 2004 09:02:02 UTC