Re: [UCR] Action 9 and discussion of UCs

Francois Bry wrote:
> Christian de Sainte Marie wrote:
> 
>>My point is that I am not sure that this is what the "XML data"
>>requiremnt is about ("RIF must be able to accept XML elements as data").
> 
> Think of practical and reasonable possible applications like the following:

[list of practical and reasonably possible applications where the data 
set against which the rules conditions are evaluated is an XML document 
or where the execution of the rules affects an XML document or both]

> More examples could be given. In   my opinion, it would be a very
> significant and very undesirable restriction if the RIF can not nbe used
> for examples like those stressed above. I therefroe dreaw the
> conclusion: "RIF must be able to accept XML elements as data".

I fully agree that there are lots of pratical and useful cases where the 
rules interchanged in RIF apply to XML element as data. Use case 1 is an 
example. I think the BPEL use case is one, too. The attached use case, 
provided by Tracy Bost of Valocity in another framework (MISMO BREW) and 
reproduced here with his permission, is yet another one.

But I am not so sure about your conclusion...

In the kind of architecture that I have in mind (see [1], for an 
example), the RIF rules are translated in the rule language used by the 
consumer application. In the above use cases, either the said rule 
language must accept XML elements as data or the said consumer 
application must translate these XML elements into whatever data 
structure the rule language accepts as data.

What I am not sure of, is how does this relates to: "RIF must be able to 
accept XML elements as data".

Actually, I am not sure to understand what that requirement means 
exactly: if it is something like allowing the vocabulary in a rule 
(predicates, functions, types, constants, whatever) to be specified in 
an agreed-on XML schema, then, yes, all the above examples support the 
requirement. But, then, I think that both the name and the wording of 
the requirement are misleading.

If, however, it is rather something like allowing the use of, e.g. XPath 
fragments to represent atoms in RIF, I do not think that the requirement 
follows from the use cases. That could of course seem like an obvious 
choice when the data on which the rules apply are XML document complying 
to an agreed-on XML schema, as in UC 1 and the MISMO use case. But it 
would be making a strong assumption on what the consumer application 
does with the rules and on how it works (or forcing it to translate back 
the XPath fragments into its own internal representation of the XML 
document).

> QED :-)

Indeed, but Quod Est still not quite Demonstratum, in my opinion :-)
Or, at least, quod still requires some more clarification...

Christian

[1] 
http://www.w3.org/2005/rules/wg/wiki/Discussion_of_UC1_-_Negociating_eBusiness_Contracts_Across_Rule_Platforms