[
Lists Home |
Date Index |
Thread Index
]
[Dare Obasanjo]
Defining a type system via the narrow lens of validation is partly the
cause of what many term the "brokenness" of W3C XML Schema and why some
of our devs and testers (of which I used to be one) have had some issues
with the XQuery type system. Watching the same mistake repeat itself on
XML-DEV even with the hindsight provided by W3C XML Schema is oddly like
watching a train wreck in slo mo.
[I had previously said]
> In turn this notion implies the possibility of a validity
> state of "well formed and correctly structured but with the
> wrong simple types somewhere".
> That would be interesting because you could still picture
> doing useful processing, and perhaps even casting the wrong
> types to the ones you were interested in.
>
[Tom P]
If you thought I mainly types in connection with validation, I did not. I
thought mainly of simple types as they would be encountered element by
element, separate from validation. However, I was thinking about how
validation had gotten mixed up with value types, and that led to my remark,
which actually, I think, is considering value types as basically separate
from validation (hence the possibility of valid structure without value type
validity).
But if you did want to check validity of value types against something-
shall we call it a schema? - you would sometimes to be able to say what
structural locations you were requiring your value types to be at. So once
you had a workable type system for values (by "value types" I mean the
equivalent of "simple types" that is, character content of elements and
character values of attributes), it would be natural to want to use it with
your schema apparatus - pluggable types, of course.
Does any of this go against what you said above?
Cheers,
Tom P
|
