Re: PROV-ISSUE-553: QNames are too restrictive as identifiers [XML Serialization]

If we use a custom identifier type to support more "scruffy" provenance,
we may have to rely on validation approaches other than XSD-based
validation.

For example, Schematron [1] is a rule-based validation approach that can
validate prov-xml traces with conditionals such as existence and uniques.
It also allows the expression of element dependencies in the
constraints--something XSD validation does not support. Note that
Schematron is also an ISO/IEC 19757 standard for Document Schema
Definition Languages (DSDL), Part 3: Rule-based validation. However, I
think it may be better to stick with more traditional (and limited)
XSD-based validation as that is more mainstream in implementations. So in
a sense, it constrains what we really want to mean as "scruffy".


Also, I've updated our PROV-XML identifiers analysis page

http://www.w3.org/2011/prov/wiki/Prov-XML_Identifiers


with some updates and additional overall analysis.


--Hook

[1] http://www.schematron.com/



On 1/29/13 9:10 AM, "Stephan Zednik" <zednis@rpi.edu> wrote:

>Thanks for the response Luc,
>
>The question of whether we want the xml validator to check the validity
>of the prefix only really came to me when we were doing the analysis and
>should be discussed by the group.
>
>My opinion is that a well-formed identifier should be required for
>"scruffy" provenance; I would prefer to not have to fall back to a
>constraints implementation to test if an identifier value is well-formed.
> 
>
>To me this means that an xml validation of the schema should check the
>well-formedness of the identifier because validated prov-xml should be
>considered valid "scruffy" provenance.  I do not know if the group has
>formally made an decisions about the relationship between xml schema
>validated prov-xml and validated "scruffy" provenance.  In the case of
>prefixed identifier values I think checking that the prefix is defined is
>integral to checking if the identifier is well-formed.  I would like to
>hear the group's thoughts on this.
>
>I would like to add to this week's agenda a quick discussion on whether
>the xml schema should check the well-formedness of the prov:id value and
>if testing the validity of a prefix is a requirement for checking the
>well-formedness of a prefixed identifier value.
>
>--Stephan
>
>On Jan 29, 2013, at 7:06 AM, Luc Moreau <L.Moreau@ecs.soton.ac.uk> wrote:
>
>> Hi Stephan,
>> 
>> If we didn't require the xml validator to check a prefix is properly
>>declared, then
>> we could define a grammar for qualified names.
>> But if we require the xml validator to check the validity of a prefix,
>>then, I agree
>> with your analysis.
>> 
>> Luc
>> 
>> On 24/01/2013 18:43, Stephan Zednik wrote:
>>> I have updated the wiki page on XML Identifiers
>>> 
>>> http://www.w3.org/2011/prov/wiki/Prov-XML_Identifiers
>>> 
>>> with additional advantages and disadvantages on using QName and anyURI
>>> 
>>> I have also added my personal opinion on the options in the Analysis
>>>section.
>>> 
>>> --Stephan
>>> 
>>> On Nov 20, 2012, at 1:28 PM, James Cheney<jcheney@inf.ed.ac.uk>  wrote:
>>> 
>>>   
>>>> Hi,
>>>> 
>>>> At the last face-to-face meeting I agreed to check with Henry
>>>>Thompson about this.  I thought there was a tracked action about this
>>>>but there doesn't seem to be.  In any case, I asked Henry and he said
>>>>he'd try to look at it this week.
>>>> 
>>>> --James
>>>> 
>>>> On Sep 13, 2012, at 3:12 PM, Provenance Working Group Issue Tracker
>>>>wrote:
>>>> 
>>>>     
>>>>> PROV-ISSUE-553: QNames are too restrictive as identifiers [XML
>>>>>Serialization]
>>>>> 
>>>>> http://www.w3.org/2011/prov/track/issues/553
>>>>> 
>>>>> Raised by: Luc Moreau
>>>>> On product: XML Serialization
>>>>> 
>>>>> 
>>>>> The current schema uses xsd:QName for identifiers.
>>>>> 
>>>>> As a result, the following identifier is not allowed:
>>>>> 
>>>>>       <prov:activity prov:id="pc1:00000p1">
>>>>>           <prov:label>align_warp 1</prov:label>
>>>>>       </prov:activity>
>>>>> 
>>>>> An xml schema  validator would complain with:
>>>>> 
>>>>> 'pc1:00000p1' is not a valid value for 'QName'
>>>>> 
>>>>> 
>>>>> Solution: allow for prov:QualifiedName (as defined in prov-n).
>>>>> 
>>>>> Luc
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>>       
>>>> 
>>>> --
>>>> The University of Edinburgh is a charitable body, registered in
>>>> Scotland, with registration number SC005336.
>>>> 
>>>> 
>>>> 
>>>>     
>>> 
>>>   
>> 
>> -- 
>> Professor Luc Moreau
>> Electronics and Computer Science   tel:   +44 23 8059 4487
>> University of Southampton          fax:   +44 23 8059 2865
>> Southampton SO17 1BJ               email: l.moreau@ecs.soton.ac.uk
>> United Kingdom                     http://www.ecs.soton.ac.uk/~lavm
>> 
>> 
>> 
>> 
>
>

Received on Thursday, 31 January 2013 09:14:28 UTC