Re: Feedback on P3P-2000-04-24

Renato,

Thank you for your comments on the 24 April
P3P last call draft.

> In Example 3.1, the use of the <DATA> element to group many of the other
> data elements seems very non-XML-ish. We would image that:
>   <DATA ref="#business.name">CatalogExample</DATA>
> make more sense as:
>   <BusinessName>CatalogExample</BusinessName>

This is mostly a matter of style. The style we chose may not
look XML-ish, but it does look RDF-ish. Also, it allows us
to attach attributes and subelements to our data elements.

> Also the use of elements for fixed vocabulary terms is
> also odd: For example:
>   <PURPOSE><custom/><develop/></PURPOSE>
> seems better as:
>   <PURPOSE>custom</PURPOSE>
>   <PURPOSE>develop</PURPOSE>

See answer above. Also, this syntax would not work for
our other-purpose (see section 3.3.3).

> Also, we suggest lowercase for all element names.

Again a matter of style. We believe the format we use
improves readability.
 
> It is also not clear why you use RDF/XML syntax for the Policy Reference 
> File
> and XML syntax for the Policy file?

We were originally going to use RDF/XML for everything, but decided
to stick with plain XML in the policy file because the RDF added
a lot of complexity, and would require the use of an RDF parser
(which would be an added burden on implementers). The policy
reference file is simple enough that using RDF does not add to
the complexity, and an RDF parser is not essential. The policy file
can be represented using an RDF data model, and we expect to
issue a W3C Note explaining the details of that.

> We are assuming that the specification of new data elements (Section 4
> Data Schemas) will be replaced/updated with XML Schema (Structures) work.
> This seems to make more sense that yet-another-schema definition.
> Also, that the XML Schema Datatypes will replace your sections 4.2 
> Primitive Data Types and 4.3.1 Dates

We expect this to happen in a future version of P3P. 

> Sections 4.3.2 Names, 4.3.4 Telephones, 4.3.5 Contact Information, and
> 4.4.1 User Data all seem to be remarkably similar to what vCard defines.
> It would seem to be more  feasible to simply use the vCard semantics
> (and their XML Namespace) in the policies instead of re-inventing them.

This issue had been raised earlier in the P3P process and we had little 
cooperation from the vCard folks (they actually told us that they didn't 
see a need for us to use vCard!). When the issue was raised again, 
we looked into using XML vCard or RDF vCard. However, we could 
not find anyone with implementation plans for XML or RDF vCard. 
There were also some concerns about the XML vCard syntax -- 
namely their use of multiple attribute values seperated by spaces. If 
we were to use this syntax we would want to change it. We decided 
there was not a compelling reason to switch to vCard, especially since 
many applications are using their own address book formats (for 
example Palm Pilots) and converting data in and out of vCard fairly easily. 

> The P3P spec establishes a metadata framework that should be
> reusable for other similar functional requirements. For example, PICS
> seems to operate in the same fashion. Has an attempt been made to
> generalise this metadata framework?
> If so, then RDF must be the major syntax carrier? Is it possible to use RDF 
> Schema?

It is probably possible to use RDF Schemas, and we will look into doing
that for future versions of P3P. Given the timing of the RDF schema
development, using them now would have added too many dependencies
to the P3P process and substantially delayed our specification.

> Also, in Section 2.2.1 some comments should be made on the use of the
> experimental RFC (HTTP Extension Framework). When is it due to be 
> formalised?
> Is this the W3C preferred way to associate metadata with a resource?

We do not know the disposition of the HTTP Extension Framework within
IETF. We are using it because it seems to do what we need to do.
But as far as we know there is no preferred way to associate metadata
with a resource.

Regards,

Lorrie Cranor
P3P Specification Working Group Chair

Received on Wednesday, 10 May 2000 22:07:23 UTC