Re: [XML-Data] review of current draft

Hi Mike,

Thanx for the detailed review: I cannot believe you found so many typos 
etc in such a short time :-)

I have corrected, all of them, I think. I just uploaded the corrected 
version.

Below, I will only comment on those of your remarks that called for more 
than just doing it...

Mike Dean wrote on 17/06/2010 14:17:47:
> 
>   Inconsistent use of . vs ; to end final sentences in list.  I 
> prefer consistent use of .

I do not know what is the correct usage in American english. I used the 
one for French, instead, that is, semi-colon at the end of all items 
except the last. That i sbecause, that way, I had to correct only two 
lines :-)

I will find out and do the right thing for the next publi.

>   The minor deviations from XDM concern me, because they may 
> preclude re-using code developed for XDM.

I added an editor's note, and I will raise an issue.

>   It's perhaps confusing that Id isn't an id, unless you're trying 
> to make a specific point.

No, I was not trying to make a point: I changed the name to "PIN".

>   The document needs some examples of RIF rules using XML data.  I'd
> start with these before delving into the semantics, which will lose 
> most readers.

I added an example, see below. I'd like somebody to check that it makes 
sense. If not, I wil lremove it...

------------start example------------
Example 4.1. In a RIF document that imports the sample XML document from 
Section 3.6. Example of a data model instance, associated with the 
corresponding XML schema, the first rule, below, says that an 
EarlyCustomer is a Customer whose Account number is lower than 1000: 
Forall ?x (_EarlyCustomer(?x) :-
           And( ?x # <http://example.org/customertable#Customer>
                Exists ?y (And ?x[<
http://example.org/customertable#Account -> ?y]
                               External(pred:numeric-less-or-equal(?y 
1000)))))

Notice that, if the XML document were imported without the XML schema, the 
RIF consumer processing the combination would only have access to the 
string value of the Account sub-element, without an indication of its 
type. In that case, to guarantee that the rule behave as expected, the 
producer of the RIF document would have to add the information that the 
value of ?y must be cast into an integer before being compared as a 
number. 
Another example, below, shows a rule that involves a combination with data 
that is represented as an attribute in the XML document.: 
Forall ?x (_EnglishRec(?x) :-
           ?x[<http://www.w3.org/XML/1998/namespace#attribute(lang)> -> 
"en"^^xs:language])

That rule could be intended to mean that, if an item is represented, in 
the imported XML data, by an element with an attribute named lang in the 
XML namespace, and the value of that attribute is the xs:language constant 
en, then the information regarding that item is recorded in english. 
------------end example--------------

>   It would be great to be able to use @NAME instead of attribute
> (NAME) for consistency with XPath, etc.

I am pretty sure that this is what I wanted to do, initially, and that 
there is a good reason why it was not a good idea, or not possible, or 
whatever.

I gave myself an action to check, or come up with a proposal (we need to 
track the discussion, but I am not sure that it is worth an issue, hence 
the action).

>   Is _Customer_John the actual symbol, or intended as a representative 
skolem?

Well, I wanted to avoid variables, so I needed an actual symbol, and I 
choose rif:local because they are easier to write than rif:iri and they do 
not have a value space either.

In the next version, I will probably use variables, instead: they are less 
confusiong!

>   Without a namespace, it would seem much more appropriate to use 
> symbols than strings for "Name", "Account", and "Customer".

But what are "symbols"? We do not have a "symbol" symbol space in RIF. Or 
did I misunderstand your point?

> Is _Customer_John # "Customer" valid RIF?

I think so: or is it a problem that a string also identifies a class?

Well, really, the problem is that the lexical space of rif:iri's is 
absolute IRIs only. But there is not much we can do about it at this point 
:-(

>   schema valid or schema valid should presumably include something 
different

Yes, of course. Do you think that the text is not explicit enough about 
that? Well, there was that editor's note already, that said that the 
semantics would be further explained and commented in a future version, so 
I did not change anything...

>   Treating RDF as XML is dangerous, since there are many ways of 
> expressing the same thing and the source serialization could change.

Maybe. But RDF/XML exists, and we have already a specification for RIF and 
RDF combinations. So, I do not think that we can escape without, at least, 
discussing the relationship between that spec and RIF+XML data...

>   RIF references should be updated from Working Draft to (Proposed) 
> Recommendation

I changed the text to Recommandation. But I do not have the URL for the 
docs, yet. I will see that with Sandro. 

Thanx again for your comments.

Cheers,

Christian

IBM
9 rue de Verdun
94253 - Gentilly cedex - FRANCE
Tel. +33 1 49 08 35 00
Fax +33 1 49 08 35 10


Sauf indication contraire ci-dessus:/ Unless stated otherwise above:
Compagnie IBM France
Siege Social : 17 avenue de l'Europe, 92275 Bois-Colombes Cedex
RCS Nanterre 552 118 465
Forme Sociale : S.A.S.
Capital Social : 611.451.766,20 ?
SIREN/SIRET : 552 118 465 03644