[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Scheme Resuest for Implementation

On Fri, 13 Nov 1998, Olin Shivers wrote:

> I have one problem with the current SRFI situation -- it seems a bit
> restrictive. There is a need for documents that do other than simply
> spec a language feature or interface. E.g., a requirements list of
> what a particular feature should have (such as a condition system),
> or other bits of encapsulated wisdom. Not all RFCs spec protocols.

This seems like a good idea to me.

> I note, for example, the SRFI-0 is really not a language-feature spec,
> although it does involve a small one. It's really mostly a procedure
> spec.

What's the difference between a language-feature spec and a procedure
spec in this context?

>  I think either
>     - SRFIs should be extended to allow these kinds of documents,
>       *at the discretion of the editors* (i.e., a dictatorial system,
>       run by a Scheme-like "Jon Postel" person, who enjoys trust and
>       respect, or perhaps a 2-out-of-3 acceptance mechanisms, given
>       3 authors)
>     - A *parallel* set of documents, SRFCs, should be established
>       for non-implementation-requesting documents.

What was the main reason for making SRFIs specific to implementation?  Why
not adopt a more RFC-style approach, allowing some SRFCs to contain, e.g.,
reference implementations?  I guess this goes back to the original goals
of SRFIs...

-- Donovan Kolbly                    (  RScheme Development Group
                                     (  d.kolbly@rscheme.org
				     (  http://www.rscheme.org/~donovan/