On Sun, 2006-05-21 at 07:47 -0700, bear wrote:
>
> On Wed, 17 May 2006, Jonathan S. Shapiro wrote:
>
> >I would like to propose that there should be some unique top-level
> >module name that is reserved for use by the implementation.
> >
> >I am inclined to suggest that an appropriate name for this top-level
> >module might be "SCHEME-RUNTIME", but I am not stuck on any particular
> >name.
> >
>
> This is such a good idea, I think it should be done twice!
> My proposals are "R6RS" which "SCHEME-RUNTIME" can be an alias
> for, and then each implementation should have another top-level
> module that they can pick their own name for, where all their
> extensions to R6RS live.

As much as I enjoy the enthusiasm, I do not agree.

1. There is no compelling reason to have more than one reserved name.

2. Bear's approach presents forward compatibility problems. This round
   it will be 'r6rs'. Next round it will be 'r6.1rs'. And so forth.

   That is, Bear is tacitly reserving an entire subspace of identifiers.

   An analogy: if we follow Bear's approach, we would not have defined
   SCHEME-REPORT-ENVIRONMENT to take an argument. We would instead have
   given it the name R5RS-REPORT-ENVIRONMENT.

I think that R5RS got this issue right.

shap