RE: XQuery Updates: Call for Use Cases

Dear Leonid

 

Here is my personal take on your comments:

 

1)       We are specifically specifying a DML and not a DDL. A DDL
heavily depends on the context of your implementation environment. If we
want XQuery to be widely used, then we should make sure that the
standard language defines what is minimally necessary and leave the rest
to the specifications of the implementation environment. For example,
the SQL-2006 standard defines how XML documents and "collections" are
being defined in relational database systems, or the different language
libraries in main memory XML processing already define how they create
and drop documents.

2)       Changing the value of a typed node can be defined without
ambiguities.

3)       You are correct that changing the parent property has its
problem (thus, we should not specify a MOVE statement), However, the
requirement document wording is not a problem. It does not say which
properties we will in the end change. The properties that we think about
at the moment are: the local name, namespace-uri, the content and the
typed value. We are aware that many implementations are using node
numbering schemes such as ORDPATHs etc, so we will be careful about
specifying which properties we will allow to be changed.

 

Best regards

Michael

 

________________________________

From: www-ql-request@w3.org [mailto:www-ql-request@w3.org] On Behalf Of
Novak
Sent: Friday, November 11, 2005 1:50 AM
To: www-ql@w3.org
Subject: Re: XQuery Updates: Call for Use Cases

 

1)   There is no specification concerning the issues of XQuery DDL
syntax and semantics so far. Are you planning to include CREATE and DROP
document (and e.t.c.) statements to the update specification or this the
subject for individual specification?

2)   The "Changing values" functionality seems to be the potential cause
of the appearance of the ambiguities during the document update: as far
as changing the typed-value of the node doesn't uniquely determines the
modification of the node' content.

3)   "The XQuery Update Facility MUST
<http://www.w3.org/TR/xquery-update-requirements/#terminology-must>  be
able to change the properties of existing nodes while preserving their
identity". I guess it is very problematically to fulfil this property in
some native XDBMS implementations, since such a properties of the node
like 'parent' property is strongly connected to the node identity (for
instance, when numbering scheme is used for node identification)

 

Regards, Leonid Novak.