Re: [sca-bindings] Another revision of proposal to resolve issue 54 - #10

[Simon Holdsworth <simon_holdsworth@uk.ibm.com>]

I've got some comments on this version
of the proposal, mostly style rather than content.  Apologies for
rehashing this yet again...

1) For the text:

A binding.ws element MUST NOT contain more
than one of the following:

• the @uri attribute
• the @wsdlElement attribute referring to a WSDL port or to a WSDL service
• the endpointReference element

Could
you interpret that as there shouldn't be more that one uri, and not more
than one wsdlElement, etc. rather than not more than one of any of them?
 I think "any combination" is clearer that "more than
one". In the interest of conformance statement writing I'd actually
prefer:

A binding.ws element MUST NOT contain
any combination of the @uri attribute, endpointReference element and @wsdlElement
attribute referring to either a WSDL port or service.

2) Having done the conformance statement
rewriting, I'm not too keen on having a MUST in the @uri part of the following
text:

For a reference binding, the SCA runtime
MUST apply the appropriate rule from the following, and honor the resulting
URI.  If none of these rules are satisfied, the SCA runtime MUST raise
an error.

• If the @uri attribute is present then it MUST be absolute; use this value for the URI.
• If the endpointReference element is present, then use the URI from the wsa:Address element contained therein
• If the @wsdlElement attribute is present

and
references a wsdl:service element, use one of the URIs for one of the contained
wsdl:port elements from the set of available ports as specified  under
the definition of the @wsdlElement attribute.

and references a wsdl:port element, then use the URI indicated by the port element

I think we should to rewrite the @uri case as:

If the @uri attribute is present and is an absolute value then use this value for the URI.

3) Should the endpointReference case be defined as having to be absolute (and if not, why not?)

4) The term "honor" seems a bit loose in a conformance statement.  The output of this determination is used at the assembly level to determine what proxies are provided to the calling application.  Is there a more specific term we can use here?  I would suggest something along the lines of "The SCA runtime MUST use the resulting URI to invoke the targetted service" but that won't necessarilty be true.

5) Following what we've done for issue 23, I would actually prefer this to be phrased using the definition of a new term along the following lines:

The target address for a reference binding is defined as one of the following:

• The value of the @uri attribute if an absolute value
• The value of the wsa:Address element of the endpointReference element (if an absolute value?)
• The value of the address element of the WSDL port referenced by the @wsdlElement attribute
• The value of the address element of one of the set of available WSDL ports as specified under the definition of the @wsdlElement attribute when it references a WSDL service element

If there is no target address for a
reference binding, the SCA runtime MUST raise an error.
For a reference binding, the SCA runtime
MUST honor the target address.  <-- don't like "honor"
here.

Here's a version of the proposal with
my changes in it, except for replacing "honor".



Regards, Simon

Simon Holdsworth
STSM, SCA Bindings Architect; Master Inventor; OASIS SCA Bindings TC Chair
MP 211, IBM UK Labs, Hursley Park, Winchester SO21 2JN, UK
Tel +44-1962-815059 (Internal 245059) Fax +44-1962-816898
Internet - Simon_Holdsworth@uk.ibm.com

Eric Johnson <eric@tibco.com>

20/05/2009 18:47

To	OASIS Bindings <sca-bindings@lists.oasis-open.org>
cc
Subject	[sca-bindings] Another revision of proposal to resolve issue 54 - #10

Attached you should find the revised proposal for issue
54 - attempt #10

Changes:

• Under "endpointReference" definition in section 2, deleted the second sentence that starts "When this element is present...", as we decided this was redundant with the new normative statement.
• Replaced reference to "runtime" with "SCA runtime" where I had left it unqualified.
• Updated the normative statement at the end of section 2 to use the bullet point construction that Anish suggested.
• Updated the first paragraph under section 2.2 with the agreed-upon text from the chat room - except that I felt it necessary to tweak the last sentence a little further.  Specifically, I came up with "For example, the service endpoint URI might be produced by modifying any or all of the host name, the port number, and a portion of the path."  What we had in the chat room didn't quite work for me.
• Updated the third paragraph in section 2.2 with the alternate text Simon N. proposed in his email.

-Eric.---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail.  Follow this link to all your TCs in OASIS at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php

---

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU

sca-binding-ws-1.1-spec-cd02-issue54-proposal-10.doc

sca-binding-ws-1.1-spec-cd02-issue54-proposal-10-sh.doc