[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: bohemians, gentry

• To: David Carlisle <davidc@n...>
• Subject: Re: bohemians, gentry
• From: Jonathan Robie <jonathan.robie@d...>
• Date: Sat, 07 Dec 2002 19:47:22 -0500
• Cc: xml-dev@l...
• In-reply-to: <200212050103.BAA01648@n...>
• References: < <5.1.0.14.0.20021204184957.01638bc0@n...><5.1.0.14.0.20021204184957.01638bc0@n...>

At 01:03 AM 12/5/2002 +0000, David Carlisle wrote:

> > As far as I can tell, XQuery and XSLT will both allow you to process data
> > in this way.
>
>No, they won't as Xquery's bias towards schema typing means that an
>implementation is allowed to just use the typed value and junk the
>original lexical string so if the original was
><Weight>123.456</Weight>
>you might get away with it but if it's
><Weight>0<!--I added this leading 0 for a reason-->123.4560</Weight>
>a query processor is allowed to report that as
><Weight>123.456</Weight>
>and information is lost.

Yes, this is true. The reason for this is that some XQuery implementations 
may be creating strongly typed environments to be stored, eg, in a 
relational database or an object - and these systems have no way of storing 
a comment in between the digits of a single decimal number.

>It is lost because Xquery views the stated
>typing as more important than the XML markup. It is essentially not an
>XML processing language at all; it is language for querying typed trees.

A processor is only allowed to canonicalize typed data, and is certainly 
never required to. In the absence of a datatype, the application could not 
safely canonicalize, but the datatype says clearly how the data is to be 
interpreted.

And one reason this can be helpful is the kind of interoperability problems 
illustrated by your example: How many numbers does the following element 
contain?

         <Weight>0<!--I added this leading 0 for a reason-->123.4560</Weight>

Of course, you could simply say that depends on the application, and the 
markup should not mandate whether one weight or two is intended...but 
doesn't that tightly couple your data to the applications used to process it?

Jonathan

• Follow-Ups:
  • Re: bohemians, gentry
    • From: Uche Ogbuji <uche.ogbuji@f...>
  • Re: bohemians, gentry
    • From: David Carlisle <davidc@n...>

• References:
  • Re: bohemians, gentry
    • From: Jonathan Robie <jonathan.robie@d...>
  • Re: bohemians, gentry
    • From: David Carlisle <davidc@n...>

• Prev by Date: Re: The privilege of XML parsing - Data types,binary XML and XML pipelines (LONG)
• Next by Date: Re: Typing and paranoia
• Previous by thread: Re: bohemians, gentry
• Next by thread: Re: bohemians, gentry
• Index(es):
  • Date
  • Thread