{Disarmed} Re: PROV-XML staged, ready for review

Hi Stephan, Curt, and Hook,

In the interest of time, I am sending this now. I will scan my copy 
tomorrow
when I am in the office next.

I like the document. It will make a good specificiation.

We are nearly there. There are a few issues that I would like to see
fixed before publication.

1. A while back, you rightly made the case that elements within a
bundle/document should not be ordered.  However, recently, you
reintroduced a sequence here (probably when trying to fix the jaxb
issue). I don't think it should stay, and it should be reverted back
to a choice: I think the generated jaxb will still be ok.

2. As part of the design section, we should talk about the identifier
conventions, especially, that they go beyond what XML does usually.

   1. All PROV identifiers are expressed as QNames in prov-xml.

   2. PROV mandates that QNames should be mappable to a uri by concatenating
      namespace uri to local name.

   3. Note the restrictive nature of QNames, and some PROV-N identifiers,
      such as ex:001 are not valid QNames.  Hence, from an inter-operability
      viewpoint across representations, implementers should aim to 
express identifiers
      as valid QNames rather than more permissive QualifiedNames in PROV-N.

3. (A previous comment that was not implemented)

     While the examples probably XML-validate, many of them do not
     satisfy the convention 2.1 above.

    for instance <agend id="ag/> require, either:
     - a default namespace to be defined xmlns="http://ex.org", or
     - a prefix to be added <agend id="ex:ag/>

    My preference is for the latter option.

    See attached file for all occurrences I have found problematic.

4. (Another comment I had raised)

    I find the use of terms "attribute", "element" and "type" very 
confusing.

    I suggest a systematic replace as follows, to distinguish the prov
    and xml usages of the terms.

    - attribute to become  prov-"attribute" or xml-attribute
    - element to become  prov-"element" or xml-element
    - type to become  prov-"type" or xml-type

    For prov, I would also put them in quote.

    Likewise, see attached file for some occurrences. But systematic
    search is required.

5. At various places, it should be clear that you are borrowing
    prov-dm definitions or figures (of provenance, or components).
    Also the introduction seems to be copied verbatim from prov-dm.


6 Can table 1 contain links to xml schema definitions?

7. I pointed out that usage of "reference" and "relation" was a bit strange.
    For reference, I suggested "denote".

    e.g.   The xml-element prov:activity is used to DENOTE a prov:Activity

    A PROV type attribute RELATION may be inferred ...

    I don't know what RELATION you mean here.

See attached file for typos, and minor issues.

Thanks,
Luc



On 04/04/2013 10:21, Stephan Zednik wrote:
> link to Overview.html 
> https://dvcs.w3.org/hg/prov/raw-file/default/xml/releases/NOTE-prov-xml-20130430/Overview.html
>
> Note - links to w3 resources such as respec.js, stylesheets, and 
> images are not using HTTPS and so will not show up browsers that do 
> not follow http links from https pages.  Please view in Firefox or 
> Safari (not Chrome) or view by opening the file in your local copy of 
> the repository.
>
> --Stephan
>
> On Apr 4, 2013, at 3:17 AM, Stephan Zednik <zednis2@rpi.edu 
> <mailto:zednis2@rpi.edu>> wrote:
>
>> I have staged the PROV-XML 2013-04-30 Note Release
>>
>> https://dvcs.w3.org/hg/prov/rev/adc235306129
>>
>> validation: http://bit.ly/ZbNuQX
>>
>> Curt, could you please generate the diff.html and add it to the 
>> staged directory?
>>
>> Also, I will not be able to be on the entirety of the call tomorrow, 
>> but I should be able to attend the first 20 mins or so.
>>
>> --Stephan
>

-- 
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