[plt-scheme] Re: Is R6RS useless for PLT?

On Tue, Nov 25, 2008 at 10:07 AM,  <kbohdan at mail.ru> wrote:
> Sam TH wrote:
>>
>> On Tue, Nov 25, 2008 at 5:37 AM, Tom Gordon
>> <thomas.gordon at fokus.fraunhofer.de> wrote:
>
> <snip>
>>>
>>> Type Scheme is a nice experimental language.  I really appreciate this
>>> effort.  But I would have appreciated it even more if it had been
>>> developed
>>> as an extension of R6RS.
>>
>> I hope you appreciate that this is not actually possible.  Please see
>> my paper with Ryan Culpepper and Matthew Flatt in the Scheme Workshop
>> last year for the implementation details, very few of which can be
>> done in R6RS.
>
> I think Tom assumes corresponding extensions to R6RS core.
>
> Standard or de facto standard for Typed Scheme would be miracle.
> And such standardizing would be much easier if Typed Scheme is
> based *mostly* on R6RS.

I really suggest you look at the paper.  Typed Scheme doesn't rely on
PLT specific extensions in the sense of an extra function here or
there.  It relies fundamentally on aspects of the PLT macro and module
system that do not have analogues in the R6RS (such as the language
position and #%module-begin).  PLT Scheme, in a sense, *is* the
extension needed to implement Typed Scheme.

I think this really gets to the heart of the question under
discussion.  Some small modules in PLT Scheme could be ported to R6RS
(scheme/bool, for example).  But virtually anything that would be
interesting relies fundamentally on the extensions that we've
developed, and that we don't want to sacrifice in the name of
portability.  We do not see PLT as simply a mechanism for developing
potential extensions to the R6RS.

Thanks,
-- 
sam th
samth at ccs.neu.edu