Re: [Scheme-reports] 5. Program Structure

On Mon, Jan 7, 2013 at 8:48 PM, Andy Wingo <wingo@x> wrote:

Hi,

I don't have much more to say about the r7rs, but there is this one
point.

On Sun 06 Jan 2013 02:53, Alex Shinn <alexshinn@x> writes:

> Also it's not true, unfortunately! Because in the next section,
> "5.3.3
> Multiple-value definitions", we see the introduction of
> `define-values'
> with an optimistically short specification and no corresponding
> expansion.
>
> Define-values is just derived syntax, it doesn't need special
> treatment.

Ah, I hadn't seen the expansion in the appendix. It does seem to be
broken, though; var0 will hold the list of values, not the first value.

I forgot to double check on this. I just did and believe you

read the reference implementation wrong. The derived syntax

comes directly from the chibi definition, which works fine and

passes the define-values case in the R7RS test suite.

Also, if used at the top-level, the above and the report expansion may
or may not bind an unrenamed variable ("dummy" or "values"), depending
on the implementation. Not so great.

This is a corner case that implementations handle differently,

but the correct behavior is to rename the binding so it's not

an issue.

Also the reference implementation has the advantage that

you only ever see the "dummy" in the degenerate, zero-values

case.

Alex