[r6rs-discuss] Re: [Formal] formal comment (ports, characters, strings, Unicode)

From: William D Clinger <will>
Date: Tue Mar 20 16:53:01 2007

I am posting this as an individual member of the Scheme
community. I am not speaking for the R6RS editors, and
this message should not be confused with the editors'
eventual formal response.

Thomas Lord wrote:
> 5.92 in referring to a "set of characters", in the context of a
> programming language, surely means that set to be constructively
> defined in all implementations and to include a primal equality
> relation (let's leave aside for a moment whether or which of
> EQ?, EQV?, or EQUAL? reflect that equality relation).

It most surely does not mean that. All previous reports
have been axiomatic in character, not constructive, and
the current draft R6RS follows that model (albeit not so
well as one might hope). The R5RS neither mentions nor
defines a "primal equality relation", and neither does
the current draft R6RS.

Since your argument rests upon this illusory foundation,
your argument is fundamentally flawed.

> The language of 5.92 describes a total order over a set, not a
> total order over equivalence classes of a set.

The notion of a set is itself relative to some notion of
equality. If you don't believe me, go look at ZF.

> No, but 5.92 refers to a constructive set with a reasonable equality
> predicate and also goes on to describe that set as totally ordered.

No, the draft states axioms that all implementations must
satisfy with respect to particular equality predicates.
It neither states nor implies anything with respect to
equality predicates that are neither mentioned nor implied
by the draft.

> It is "lawyerly" in the worst possible sense of the word to argue,
> retrospectively, that really two different notions of equality are
> being used.

The draft itself refers to several different notions of
equality, and is usually careful (though sometimes not as
careful as it should be) to identify those predicates.

That is common sense.

For you to hold that the draft states or implies that
char<=? is a total ordering with respect to arbitrary
implementation-dependent predicates, after I have written
a concrete example to illustrate the absurdity of such a
claim---well, that seems pretty disingenuous to me.
There is no need for either of us to insult the lawyers.

Will
Received on Tue Mar 20 2007 - 16:52:54 UTC