RE: WS-Addr issues

Gudge wrote on 11/06/2004 04:07:10 AM:
...
> RefProps/Params are NOT a generic way for specifying what SOAP headers a
> service requires. They *ARE* part of the address information. Again, to
> use the postal analogy, consider the address:
> 
> Financial Controller
> Department 63465
> One Microsoft Way
> Redmond
> WA
> 12345
> 
> The postal service will use the last four lines (at most) to get the
> letter to the correct place. The internal postal delivery service inside
> Microsoft will use the top two lines to get the letter to the correct
> person. I see the former as being like the [address] portion of an EPR
> and the latter as being like [reference properties]. Both are needed to
> get the letter into the right hands.

The problem with this analogy is that each intermediary along the way
knows exactly where to look for each piece of data.  They may only know
how to interpret their own bit of info and that's ok, but they know that
its part of a bigger piece of data - the "To".  The entire address
is considered to be a single unit with sub-pieces of information useful
to different parties.  Imagine if you put the zipcode on the back of the
envelope, the state initials on the top right corner (normally where the
>From goes)...  All of the same information is there but because they are
not all grouped together into a single "To" we can't be sure all of it 
is really part of the "To".

Some may say that's ok because the receiver of the message knows what to
look for since each piece of data is wrapped by some element.  True, but
this makes (IMO) the wrong assumption that all intermediaries will know
about all of those elements.  Lets say there's an intermediary whose job 
it is modify rather than read the headers, maybe it wants to encrypt 
everything but the WSAddr info, or perhaps it wants to log the entire "To" 

for some reason w/o having to save all of the headers.  Having a 
standard/well-defined location that everyone can count on for the entire 
target endpoint could only help.

-Dug

Received on Saturday, 6 November 2004 13:43:11 UTC