Re: Personal reply to Edd Dumbill's XML Hack Article wrt W3C XML Schema

["W. E. Perry" <wperry@fiduciary.com>]

Jonathan Borden wrote:

> Silly question but it underlies something cool about XML Schema datatyping
> in that it is _possible_ to define a datatype in terms of a regular
> expression ... so the <value>45.67</value> is plain and simple text ... XML
> is just text. The fact that "45.67" matches a regexp labelled
> "floating-point" perhaps means that the call:
>
>   float createIEEEwhateverFromString("45.67")
>
> will return something useful. But the value is plain and simple text. I
> think that its very important that datatypes can be defined in terms of a
> text pattern rather than binary goo -- it allows me to tolerate the
> existence of terms like "unsigned short integer" in an XML spec.

Amen. As a tool for processing the instance, this is a delight. I never said
that the tools weren't useful, only that giving them a privileged position
necessarily creates canonical semantics, and takes us straight back to the
instance-follows-content-model bondage under which we were sore oppressed
before well-formedness liberated us.

Respectfully,

Walter Perry