Re: ISSUE-595: Prov-xml subtyping needs to be marked in the document

On Feb 25, 2013, at 9:07 AM, Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk> wrote:

> On Mon, Feb 25, 2013 at 2:54 PM, Stephan Zednik <zednis@rpi.edu> wrote:
>> The implication is that xsi:type can be used to specify the type of an element for validation purposes
>> 
>> http://www.w3.org/TR/xmlschema-1/#xsi_type
>> 
>> This is defined by XML Schema and is not something we are adding extra meaning to.
> 
> So it does not also append prov:type 'ex:Workflow' to the PROV-DM
> statements loaded from that PROV-XML?

It could.  I think that is a reasonable assertion to make, and one that we should add to the note.

> 
>> What do you mean technically by "don't understand my schema"?
> 
> I mean a PROV consumer who can see and load your schema, but has no
> programmatic understanding of anything beyond PROV-XML. Let's say this
> consumer is to output PROV-N.
> 
> 
>> For the following xml snippet:
>> 
>> <prov:plan prov:id="foo" xsi:type="ex:Workflow" />
>> 
>> If the schema that defines ex:Workflow is not known (namespace "ex" is not defined) the xml will not validate because the namespace "ex" is not defined.
> 
> (..)
> 
>> If the namespace/schema are known and ex:Workflow is an extension of prov:Plan than the xml will validate.
> 
> I am less concerned now about the XML validation; as that's known
> territory. What I wonder about is the semantics of using an XML
> extension of a complex type and/or xsi:type.
> 
> 
> So let's say we have:
> 
> <prov:entity prov:id="foo" xsi:type="ex:Workflow" />
> 
> and in the schema the complex type ex:Workflow extends the prov:Plan
> complex type.
> 
> Let's say the PROV-XML consumer can fully load the schema and the XML
> validates perfectly, and then saves the PROV statements as PROV-N.
> 
> 
> Would you expect then to find in the PROV-N:
> 
>  entity(foo, prov:type='ex:Workflow')
> 
> and more importantly the inferred:
> 
>  entity(foo, prov:type='prov:Plan')

A good question and I think the PROV-N encoding should include the these statements.

I can add text to reflect these type inferences to the note.

--Stephan

> 
> 
> ?
> 
> 
> --
> Stian Soiland-Reyes, myGrid team
> School of Computer Science
> The University of Manchester
> 

Received on Monday, 25 February 2013 16:23:51 UTC