[I'm resending this because the "From" bug truncated my previous
post.  In the original, "-From-" was "From".]

John Cowan wrote:

> Shiro Kawai scripsit:
>
> > Oops, I'm probably lost... Does the current discussion suggests that
> > fresh location tag must be allocated for every time lambda expression
> > is evaluated, even the lambda expression doesn't close any mutable
> > state?
>
> That is what R5RS requires, yes, as well as Will's proposal.

No.  John Cowan is wrong about what the R5RS requires, and he
is wrong in thinking my proposal requires that as well.  With
regard to the R5RS semantics, John has explained his reasoning.
-From- John's explanation, I know he jumped to a false conclusion
via an invalid generalization based on his incorrect guess about
the purpose of two examples.  I will post on this at length when
I have time this evening (and access to my archives).

In addition, I'd like to warn that most of the recent speculation
about which optimizations are envisioned/affected/allowed by the
various semantics has been off the mark.  I will post about this
at length when I have time, no later than this evening.

One thing that's become clear from this discussion is that a lot
of people have been writing code that is not at all portable (in
theory) under the R5RS semantics.  In my posts later today, I will
discuss what implementors can do (and, in many cases, have already
been doing) to make this erroneous legacy code work in spite of
itself.

Will

_______________________________________________
Scheme-reports mailing list
Scheme-reports@scheme-reports.org
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports