Editorial comments on Primer

I am extremely pleased in how Parts I and II have isolated the encoding
of graphs as being distinct and independent from RPC.  Moreover, the
fact that roughly 95% of the Parts I and II are independent of encoding
or RPC is a fantastic advance, as object serialization/RPC have always
been but one application of SOAP. 

 

Unfortunately, the Primer does not seem to mirror the good work done by
Parts I and II.

 

I am concerned that the Primer reinforces the myth that there are in
fact two (and only two) distinct worlds: document and rpc. It does this
by (a) calling out non-RPC applications by name (document), (b)
discussing RPC-specifics too early and (c) intermingling RPC-isms in
other broad/general concepts. Moreover, the specific examples seemed to
have a very artificial distinction.

 

Granted, the primer needs to address RPC-specifics such as <rpc:result>
and the convention of encoding the operation/method name as the QName of
the first child of Body. I'm all for that!

 

However, given that less than 5% of the normative specs even discuss RPC
(1842 words out of a total of 37042), it felt very strange that RPC
wound up permeating the primer as it did.

 

Also, there seems to be factoring issues, in that there is a co-mingling
of the presentation of broad SOAP features inside the RPC sections.
Specifically, in section 2.2.2 of the primer, the paragraph beginning
with "RPCs may also require additional information..." as this issue is
NOT specific to RPC but rather to any app that needs to augment the
schema/content of the body with additional contextual goo.

 

DB