campbell@xxxxxxxxxxxxxxxxxxxxxxxxxxx wrote:
On Mon, 30 Aug 2004, felix wrote:Yes, so what prevents `require-extension' to be such a disjoint config-language form? If I understand correctly merely requiring that `require-extension' should be the first form in the program text (directly, not as an expansion of another macro) would make it compatible with Scheme48, right?Yes, although I don't understand why you'd want to so dramatically move away from SRFI 7 when it is such a trivial amendment to SRFI 7, SRFI 7 provides much more quite useful functionality anyway, and SRFI 7 permits extension considerably more easily: the PROGRAM form is a quite general form, whereas REQUIRE-EXTENSION has only one purpose, to declare a dependency on a feature.
The fact that `require-extension' serves a single purpose is clear and fully intended. It also appears to me more natural and straightforward than shoehorning SRFI-7 into a more usable form. If you think SRFI-7 should be amended and/or extended, I would find it very reasonable to do so. Yet, I think SRFI-55 expresses the notion of "loading/linking whatever this implementation needs to support SRFI X" in a clearer way. It might also be a good idea to recommend putting any `require-extension' forms at the top of a source-file, preceding any other top-level forms, which in the end would make it even compatible (in whatever way, I don't know much about S48's implementation) with Schemes that separate the configuration language from the actual code. cheers, felix