Shiro Kawai wrote:
>> Aubrey Jaffer wrote:
>>> I think having SRFI-101 shadow R6RS identifiers is the most common way
>>> that SRFI-101 would be used.
>
> Is this true?  Since srfi-101's data structure is incompatible
> to the built-in pairs, you can't pass srfi-101 list to the
> imported procedures that expects built-in list, and vice versa.
> (Or am I missing something?)

I don't think you're missing anything.  Passing a random-access list to
a procedure expecting a sequential list will not work.

> If you import srfi-101 shadowing r6rs list primitives, you
> pretty much commit to live in srfi-101 world, isolated from
> other libraries that deal with native lists.   That's ok
> for experimental work, but for practical applications
> I'd rather imagine that it'd be more common to import srfi-101
> procedures with prefix, and the user converts normal lists
> and srfi-101 lists as needed.

Agreed, I think this will be the more common usage.

I would like to accommodate the other usage by providing a replacement
(rnrs base) library---as a convenience only---that uses random-access
pairs in place of traditional pairs.  I am not entirely sure this can be
done as an R6RS library, but I'm not entirely sure it can't either.

David