Re: XML Schema response to XML Query Requirements

in
http://lists.w3.org/Archives/Public/www-xml-query-comments/2000Mar/0003.html
David Beech wrote:

> On behalf of the chairs of the XML Schema WG, I am pleased to
> forward our response to the XML Query Requirements WD, as
> agreed by the Schema WG today.
>
> Respectfully submitted,
>
>   David Beech
>
>
> XML Schema WG response to XML Query Requirements WD [1]
> -------------------------------------------------------
>
> XML Schema WG thanks XML Query WG for the opportunity to review
> this requirements document.  We apologize for some delay in our
> response due to our preoccupation with completing our own
> Last Call document prior to seeking to go to CR.
>
> We approve of the general scope of the requirements, and the
> reliance on information provided by XML Schema processing
> where available.
>
> We have the following comments and questions.
>
>
> 1. Usage scenarios (2.7)
> ------------------------
>
> Could some more explicit mention be made of querying XML Schemas
> in usage scenarios such as 2.7?

Response: The XML Query WG agrees. The next version
of the requirements document will refer explicitly to XML Schema
in Usage scnario "2.8 Catalog Search" in the first sentence:
"Perform queries to search catalogs that describe document servers,
XML document types, XML schemas, or documents."


>
>
> 2. Reliance on XML Information Set (3.3.1)
> ------------------------------------------
>
> We would welcome early coordination to ensure that the
> PostSchemaValidation-Infoset provides the desired information
> for the Query Data Model.

Response: The XML Query WG agrees. As also indicated in requirement 3.3.1,
the group carefully coordinates the XML Query Data Model with the
PostSchemaValidation-Infoset, and has already sent request for clarification
on
open
PostSschemaValidation-Infoset issues to the XML Schema WG
(see also:
http://lists.w3.org/Archives/Public/www-xml-schema-comments/2000JanMar/0217.
html).


>
>
> 3. Collections (3.3.3)
> ----------------------
>
> How do you intend to handle "collections of ... [instances of]
> simple and complex types" in the Query Data Model?
> Is the limited treatment in the current PSV-Infoset
> (whitespace-separated lists in text-based simple types,
> and maxOccurs for elements) a satisfactory base on which to build?

Response: No this limited treatment of collections in the
current PSV-Infoset does not suffice. The XML Query WG
anticipates that the query language will need to distinguish between
single information items and collections of infomation items, both
ordered (similar to lists) and unordered (similar to bags or sets).
The XML Query WG believes that this requires a more direct support
for collection types than is provided
by minOccurs and maxOccurs. More precisely, this raises the
following two issues:

 - Distinction between singleton collection and a single information
 item. In the case where minOccurs and maxOccurs are both 1, it
 is not clear whether the corresponding value should be treated as
 a singleton collection or a single information item.

 - Support for unordered collections. Unordered collections, such as
 bags and sets might arise during the evaluation of queries. There is
 no support for such collections in XML Schema.


>
>
>
> 4. Schema Availability (3.3.5)
> ------------------------------
>
> We suggest an additional requirement, that it should be possible
> in an XML query to specify the schema(s) to be applied to the
> input Infoset, either
> (a) to add PSV information if it does not have any, or
> (b) to override the PSV information from a previous validation.

Response: The XML Query WG thinks that this issue should be
handled by the XML Coordination Group or by XML Schema, because
the ability to specify the schema(s) to be applied to the
input Infoset is needed also for the DOM, for XSL, and possibly
for other working groups).
>
>
> 5. Operations on Schemas (3.4.15)
> ---------------------------------
>
> Please clarify the nature of access required to "the XML schema
> or DTD for a document, if there is one."  We are in process
> of determining whether the PSV-infoset for an instance
> document should contain an item referring to its schema,
> and, if so, what form this should take.
>

Response: Like the companion requirement 3.4.14 (Operations on Names),
requirement 3.4.15 (Operations on Schemas) is intended for
queries which need to treat metadata (such as the name or the type of an
element)
as data. For example, for instances of the following XML-schema fragment:

 <xsd:complexType name="PurchaseOrderType">
                <xsd:element name="shipTo" type="ShippingAddress"/>
                <xsd:element name="billTo" type="BillingAddress"/>
</xsd:complexType>

a query may need to retrieve all elements of type "ShippingAddress" or
also for all elements of all types derived from "Address".
To allow for such queries the PSV-Infoset (and accordingly the
XML Query Data Model) should explicitly model the type
of any information item, including cases where the type is
only available as anonymous type.

>
> 6. References (non-normative)
> -----------------------------
>
> XML Schema might usefully be added as a non-normative reference
> in any revision of the Query Requirements WD.


Response: The XML Query WG agrees. The next revision will include XML Schema
as a non-normative reference.

Received on Tuesday, 30 May 2000 03:16:59 UTC