SimpleCPA [was RE: [ebxml-cppa] Issue: should BusinessTransactionCharact

[Nothing to do with BusinessTransactionCharacteristics, but responding to "SimpleCPA" suggestion]

W3C XInclude (http://www.w3.org/TR/xinclude/) could support a lot of this without requiring incompatible extensions or CPA-specific functionality.

It would ease hand-editing CPAs.

The least human-reable parts of CPA, and among the ones most likely to change when building a CPA from a template, are the XML Dsig public key information elements. XInclude would allow one to write something like (exported certificate file name based on SerialNumber):

<tns:Certificate
tns:certId="P1_SigningCert">

<xi:include href="../Certs/192466156709589705231692271946384786091.xml" />

</tns:Certificate>

Or even (exported certificate file name based on IssuerName):

<tns:Certificate
tns:certId="P1_SigningCert">

<xi:include href="../Certs/CN=Partner A,O=Partner A.xml" />

</tns:Certificate>

This would alleviate the CPA author from having to copy/paste exported signature information in CPA files.

This assumes the CPA import tool resolves the XInclude statements prior to compiling the CPA information to its internal data structures.

Flattening, aka simplifying the CPPA / CPA would be a huge plus (SimpleCPA?!).

Even though people use editors to build these things - making them easier to visualize would definately aid adoption and use.

The notion of taking two CPP's and gluing them together somehow to make a CPA is nevertheless fraught IMHO. It's never really worked well - in part because of all the gnarly refID constructs to name one challenge.

Removing the BusinessTransactionCharacteristics completely would definately make it easier to align just the communications part of the arrangement. You'd still need to retrofit your business transactions - but as you note - if you referenced a BPSS instead - that would provide the road map to those.

Also - establishing the notion of profiles and templates for common configuration sets, is another key need.

In fact - if we could somehow break the CPA down into more sub-atomic parts - that could be included....and also therefore making things more pluggable...

And I completely agree that retain 100% V2.0 backward-ness is not a critical factor.

Just brainstorming here - and being completely radical - I'd like to see something like this:

and of course each template would be reflect the original CPA syntax and stuff relating to each part.

-------- Original Message --------
Subject: [ebxml-cppa] Issue: should BusinessTransactionCharacteristics
be made zero or one in future CPPA schemas (v 2.1 and 3.x)
From: "Dale Moberg" <dmoberg@cyclonecommerce.com>
Date: Tue, May 23, 2006 5:15 pm
To: "OASIS ebXML CPPA TC" <ebxml-cppa@lists.oasis-open.org>

Many months ago several people (especially Hima) noted that BusinessTransactionCharacteristics is sometimes unnecessary and in general can complicate the checks for an acceptable compatibility between CPPs when forming a CPA.

When using BPSS (especially 2.0), and when not changing any values from those in the BPSS instance, the BusinessTransactionCharacteristics attributes end up repeating information in the BPSS instance. In addition, the QOS parameters needed for a message service are generally independently documented in specialized sections of the DocExchange element or in the Transport, so the “abstract” features of the BTC tend to just summarize the real configuration details.

Since we are trying to wrap up changes, errata, and ebMS 3.0 support in the CPPA specification, I would like the TC to review this optionality issue and decide whether we should change the cardinality to allow omission of the element when it is not really needed. (and document when that is).

I recall this issue was raised when considering how to flatten the XML hierarchy of CPPs and CPAs (which most agree would make it simpler to read if not to use!). I think flattening could be done but it would be a departure from conserving the structure of CPPA 2.0 instances. Since CPPA instances are probably headed toward being something that are never “seen” in editors, but only imported and exported by software, flattening is not something I have heard much about lately. If you think we should reconsider this for the transition to committee draft from editor draft, please discuss on the list.

ebxml-cppa message