- From: <Noah_Mendelsohn@lotus.com>
- Date: Tue, 27 Feb 2001 13:29:16 -0500
- To: "Jeff Rafter" <jeffrafter@definedweb.com>
- Cc: "Henry S. Thompson" <ht@cogsci.ed.ac.uk>, "Jason Diamond" <jason@injektilo.org>, xmlschema-dev@w3.org
The principle goal for the organization of the components in the specification was ot meet the need for a rigorous and reasonably compact representation of the information needed (not the specific data structures!) to represent a schema. The sort of aggressive polymorphism used in the specification makes it more compact. With all due respect, you are presuming a goal for the schema components which is in fact a non-goal (or at best a nice to have). While there are surely cases in which you can do straightforward mappings of the abstract components to data structures in one programming language or another, that is not a goal for the design of the components, nor is it necessarily a good idea in all cases even. High performance implementations may well want to use quite different representations of the same information even if a straight mapping is possible. Surely I would expect that components like model group, particle, etc. will often disappear into compiled state machines, or the like. On the other hand, in the many cases where a straightforward mapping meets some particular need, go for it. By the way, several early users commented positively on the degree to which, in spite of it being a non-goal, the components were effective in helping to organize an attack on the fundamental problem of parsing schema documents: figuring out how to organize the information in a useful way. ------------------------------------------------------------------------ Noah Mendelsohn Voice: 1-617-693-4036 Lotus Development Corp. Fax: 1-617-693-8676 One Rogers Street Cambridge, MA 02142 ------------------------------------------------------------------------
Received on Tuesday, 27 February 2001 13:42:51 UTC