[r6rs-discuss] Re: Plausible list problems

From: William D Clinger <will>
Date: Tue Oct 3 21:38:58 2006

I am posting this as an individual member of the Scheme
community. I am not speaking for the R6RS editors.

Andre van Tonder wrote:
> Understood. However, the following statements in the document are in
> conflict, either directly or in spirit, with these statements:
>
> p. 123:
>
> "Consequently, procedures like length are only required to
> confirm that a list argument is a plausible list."
>
> - undecidable

I can only repeat that, although plausible list-ness is
undecidable, the quoted wording says exactly the right
thing, albeit in language that should be made clearer
for the non-logicians among us: Implementations are
required to confirm (i.e. prove) that the alleged list
argument is a plausible list, and must raise an exception
if that confirmation (i.e. attempted proof) fails.

That the implementation's attempt to confirm can fail
even when the argument really is a plausible list is
clear to both of us, since you and I agree that the
question is undecidable.

> p. 124:
>
> "This section clarifies the domains of procedures in the base
> library and the (r6rs lists) library."
> "These are clarifications to the domains of the procedures of
> the base library described in sections 9.12, 9.15, and 9.18:"
>
> - misleading, since what is being clarified is not the
> domain, but some superset of the domain (e.g., the domain
> of map is some unspecified subset of the set of plausible
> lists).

I agree that the quoted wording is misleading. As I
have said, I consider this to be just one example of
a pervasive problem with the draft R6RS, for which I
intend to submit a formal comment.

> p. 125:
>
> "These are clarifications to the domains of the procedures
> of the (r6rs lists) library described in chapter 12:"
>
> - ditto

Ditto.

Will
Received on Tue Oct 03 2006 - 21:38:54 UTC