[r6rs-discuss] [Formal] Requirement to detect circular lists

From: Felix <pfr6rs>
Date: Thu Oct 5 15:43:27 2006

On Oct 5, 2006, at 2:02 PM, William D Clinger wrote:

> I do not believe an illusion of safety can justify the
> inefficiency, complexity, and indeterminate exceptions
> that result from requiring implementations to raise an
> exception whenever the list arguments of a higher order
> procedure are not lists, or do not have the same length.

Just to get the question out into the open:

How do the editors, developers and implementors feel about loosening
the spec to allowing (but not _requiring_) implementations to throw
an exception if the list arguments are not lists [of the same length]
any point during the dynamic extent of an invocation?

This would give the implementations freedom to experiment with
different detection strategies. (Based on the discussions we've seen
on this list, the investigation of such strategies strikes me as a
research topic, not a solved problem.)

This spec would not be as tight as some would like, since one would
not be able to rely upon such exceptions being thrown. But on the
other hand, it would seem to simplify the specification of these
procedures.

-Felix
Received on Thu Oct 05 2006 - 15:43:27 UTC