Peter,
Your comments about MEPs with respect to applying Postel's law are
well taken, and of course are at the heart of what WSDL is about
(however imperfectly). What I had in mind was a little less
far-reaching: operation name overloading. If we applied Postel's law,
BPEL would happily consume web services that employ overloading, but
ban the practice for services offered by BPEL engines.
Postel's law is much more applicable at the WSDL level, since that
is what must be exposed between partners. It is even more applicable to
the binding/transport level, where we already have a history of
incompatible implementations in the WS context (ie, SOAP 1.1).
-Ron
Furniss, Peter wrote:
Message
I would strongly support Postel's law in
general, but I'm not sure it applies so easily to BPEL.
The detailed content and format of
particular messages sent from (be conservative) or received by (be
liberal) a BPEL process are obviously subject to Postel's law, but
that's a matter for the binding and the involved protocols, and their
profiles.
The equivalent for BPEL would first appear
to be, as Ron says, be liberal in the web services you use,
conservative in the ones you offer. However, the use/offer polarity is
determined by the WSDL MEP's in use. In the case
of asynchronous relationships, where there are paired port types
one-way operations, there isn't any obvious distinction as to which
side is doing the specifying and which is obediently following it. Even
for request/response patterns, it isn't invariably the case that the
server has freedom to choose (and thus to obey BP 1.0) and the client
has to follow the dictates of the server (whether it obeys BP 1.0 or
not), which is the basis of Postel's law.
For BPEL, or indeed for business protocols
in general, which side has the initiative is probably determined by
other features (which side has more economic clout or more deployed
implementation or less intelligent developers or got there first).
Postel's law applies as something like "be generous in what you demand
of others, providing there is unambiguity; be precise in how you follow
the demands of others, even if it doesn't seem entirely reasonable".
Peter
-----Original Message-----
From: Ron Ten-Hove [mailto:]
Sent: 01 October 2003 21:48
To: Eckenfels. Bernd
Cc: BPEL OASIS
Subject: Re: [wsbpel] Issue 47 and WS-I BP 1.0
Perhaps we should invoke Postel's law, with respect to basic
profile conformance. To promote interoperability, we should be as
permissive as possible when dealing with (other) web services, but as
conformant as possible when creating our own (BPEL engine provided) web
services.
I realize this falls mostly on implementations to worry about, but
the basic profile does contain a lot of restrictions and clarifications
for the "abstract" service model that may be of interest.. The ban on
operation overloading might be slightly simplifying, for example.
Anyway, a quick scan of the BP didn't reveal anything scary to me.
It seems to me that WS-I BP 1.0 compliance will be an implementation
worry, but as a technology that will rely on interoperation of our SOAP
infrastructure, we ought to encourage (but not mandate) BP compliance.
-Ron
Eckenfels. Bernd wrote:
Hello Ugo,
I think it is not good to speak about
violation if an implementation is able to communicate with an
larger number of systems.
Perhaps we should chnage this: "conformant
BPEL engines must be able to provide and consume Web Services according
to BP 1.0a specification. Engine should not provide Web Services with
HTTP SOAP binding other than BP1.0 conformant.
Actually I see the point of requiring BP1.0
support as a conformance rule, but I dont see a point of restricting
engines to that, even if it applies.
Mit freundlichen Grüßen
Bernd Eckenfels
Chief Architect
--
SEEBURGER AG - Edisonstr.1 , D-75015 Bretten, Germany
Fax: +49 (0)7252 96-2400 - Phone: +49 (0)7252 96-1256
mailto: - http://www.seeburger.de
-----Original Message-----
From: Ugo Corda [mailto:]
Sent: Wednesday, October 01, 2003 8:28 PM
To: Satish Thatte; Prasad Yendluri; Eckenfels. Bernd
Cc: BPEL OASIS
Subject: RE: [wsbpel] Issue 47 and WS-I BP 1.0
Right, RPC literal would be fine, but RPC encoded would be in
violation.
-----Original Message-----
From: Satish Thatte [mailto:]
Sent: Wednesday, October 01, 2003 11:23 AM
To: Ugo Corda; Prasad Yendluri; Eckenfels. Bernd
Cc: BPEL OASIS
Subject: RE: [wsbpel] Issue 47 and WS-I BP 1.0
So for
instance the RPC encoded services bound to SOAP/HTTP would be in the
“in scope but in violation” category?
From: Ugo
Corda [mailto:]
Sent:
Wednesday, October 01, 2003 11:18 AM
To: Satish Thatte; Prasad Yendluri; Eckenfels.
Bernd
Cc: BPEL OASIS
Subject: RE:
[wsbpel] Issue 47 and WS-I BP 1.0
Let me
clarify point 4 (sorry, I mislabeled it as 3) in relation to point 1.
I
think we should distinguish services that are not compliant with BP 1.0
from those that are simply out of scope for BP 1.0.
If I
have a Web service that is not bound to SOAP/HTTP, then I would say it
is out of scope for BP 1.0, so it's OK for BPEL to interact with it.
My
point 4 is about services that are within the scope of BP 1.0 and still
do not comply with its requirements.
Ugo
-----Original
Message-----
From: Satish Thatte [mailto:]
Sent:
Wednesday, October 01, 2003 11:09 AM
To: Ugo
Corda; Prasad Yendluri
Cc: BPEL
OASIS
Subject: RE:
[wsbpel] Issue 47 and WS-I BP 1.0
I doubt that
we can mandate BPEL to be used with BP 1.0 compliant services only
especially given the answer to 1 assuming it is correct, and given that
there are many services today that are not compliant (e.g., RPC encoded
ones).
From: Ugo
Corda [mailto:]
Sent:
Wednesday, October 01, 2003 10:55 AM
To: Satish Thatte; Prasad Yendluri
Cc: BPEL
OASIS
Subject: RE:
[wsbpel] Issue 47 and WS-I BP 1.0
I see
a few separate issues/questions connected to the relationship of BP 1.0
and BPEL.
1- Would
BP 1.0 be restricting BPEL to the point that some of BPEL's
functionality would not be available?
I
cannot think of any such restriction off the top of my head.
2-
Would the fact that BP 1.0 only addresses the SOAP/HTTP binding imply
that also BPEL should be limited to that type of binding?
I
don't think that anybody would imply that.
3-
Should a BPEL process be offered as a Web service that is BP 1.0
compliant?
My
answer would be yes.
3-
Would it be fair to limit BPEL use to interacting with BP 1.0 compliant
Web services only?
My
personal answer would be yes. But I am a member of WS-I, and I
understand other people might have different answers.
Ugo
-----Original
Message-----
From: Satish Thatte [mailto:]
Sent:
Wednesday, October 01, 2003 10:39 AM
To: Prasad
Yendluri
Cc: BPEL
OASIS
Subject:
RE: [wsbpel] Issue 47 and WS-I BP 1.0
For the
benefit of the non-expert could post a salient example please?
Specifically, a BPEL usage pattern that would not work if BP 1.0 is
followed but would work if any WSDL 1.1 portType is allowed. In other
words, is BP 1.0 a restriction on the WSDL artifacts we use or
potentially on BPEL itself?
From:
Prasad Yendluri [mailto:]
Sent:
Wednesday, October 01, 2003 10:11 AM
To: Satish Thatte
Cc: BPEL
OASIS
Subject:
Re: [wsbpel] Issue 47 and WS-I BP 1.0
The sections 5.5 and 5.6 in the basic profile
(http://www.ws-i.org/Profiles/Basic/2003-08/BasicProfile-1.0a.htm)
are devoted to binding aspects but, several major sections including
section 4, other sections of 5 address abstract aspects of WSDL, which
is a pretty large portion. All those are applicable BPEL IMO.
Prasad
Satish Thatte
wrote:
Most of
the BP 1.0 directives are binding related. BP also forbids outbound
operations which BPEL does not use. Can someone identify a directive
in BP 1.0 that actually affects BPEL?
Satish
