[R6RS] `eqv?' on immutable records

[R. Kent Dybvig]

> You are proposing to forbid implementations in
> which that special case of the beta rule holds.
> That would most definitely be a change from the
> R5RS.

Fair enough, although I view it as a clarification or nailing down of an
ambiguous situation, rather than a change.

> > My suggestion would inhibit compilers from making certain transformations
> > or require them to be more careful in doing so.
>
> I don't understand how forbidding both compilers
> and programmers to rely on an important special
> case of the beta rule is going to help anyone.

It would be no help to the compiler, for sure, but the programmer would be
given a consistent semantic model that holds across implementations.  My
additional proposal to include location tags for all record instances
would allow programmers to generate sets of distinct objects (magic
cookies) without requiring them to include dummy mutable fields.

Kent