[
Lists Home |
Date Index |
Thread Index
]
On Sat, 2002-10-19 at 10:14, Sean McGrath wrote:
> The theoretical expressibility of constraints in either grammer or rule
> based systems in not the central concern here. The concern is how
> robust the two approaches are in the face of (inevitable) real world
> change.
I find that neither of the two approaches (or both approaches combined)
make schema evolution easy. As Rick rightly says, the entropy in a
schema tends to increase over time.
One major problem is how a change in the schema impacts the instance
documents. If the schema changes, old documents might not be valid
against the new schema. Let alone old software libraries.
It is possible to change the schema retaining "backward compatibility".
In general this happens if new constraints only apply to new structures
and that existing constraints are either unchanged or made looser. This
is a recipe for a nice tag soup ;-)
BTW verifying all this (automatically) is not trivial especially with
schema languages like XML Schema, which allows to do the same thing in
so many ways.
It's probably easier to add/remove/change rule-based constraints, but,
again, it is not easy to check the impact instance documents. I do not
know of any tool that supports explicitly (or at all) schema evolution.
-daniel
--
______________________________________________________________________
Daniel Dui d.dui@cs.ucl.ac.uk
Dept. of Computer Science (+44) 020 7679 7192
University College London http://www.cs.ucl.ac.uk/staff/d.dui
|