xml-dev - RE: Realistic proposals to the W3C?

[Jonathan.Robie@SoftwareAG-USA.com]

Title: RE: Realistic proposals to the W3C?

Clark C. Evans wrote:

> 1.  Software implements a process (that given particular
>     state and input produces a particular output).
>
> 2.  The goal of the specification is to describe this
>     specific process so that it can be automated by
>     a machine _and_ understood by a human.

There's an assumption here that I do not share. W3C specifications are not always specifications of programs. In fact, one of my concerns about the W3C XML Schema Structures specification is that it reads too much like an English description of a computer program - in other words, it is not declarative. Non-declarative is not a problem if you want to do the same thing that the program description does, but most of our specifications are used for purposes that were not foreseen.

Schema, for instance, has become the type system for W3C XML Schema, and a formal model that accounts for type in a declarative way is the one thing that is vitally needed for this. This is not a description of a program, though programs can be written to verify various aspects of the declarative model.

Schema is also very important for mapping among sources in middleware systems, but this is not closely related to the validation task that occupies so much of the content of Schema Structures.

"Go Declarative!"

> A.  A rigourous machine description

Change that to "a declarative formal model", and I'll be happy. And you can write programs to verify the model in languages like Haskell, if you use the right formalisms.

This is the approach Phil Wadler takes, and I think it is helpful.

> B.  A fuzzy human description

For all those fuzzy humans.

Jonathan