I understand the intention of srfi-226 reference impl, but it wouldn't harm to add instructions to try it (and probably Daphne's snippet) as a comment at the beginning of control-features.sls. What do you guys think?
On Wed, Jan 8, 2025 at 6:42 AM Marc Nieper-Wißkirchen <marc.nie...@gmail.com> wrote: > The sample implementation was never meant as an actual implementation > of the libraries in this SRFI but as a proof of concept showing how to > implement all the primitives on a small set of subforms. The sample > implementation's library name does not coincide with any name that is > defined in the SRFI text. The included testing code shows how to use > and test the sample implementation. > > While I agree that a sample implementation should be a portable > implementation if possible (which is not possible in case of SRFI 226, > but see the PS below), I have always had the opinion that sample > implementations should almost never be used or included in Scheme > implementations as-is but should be tailored to each Scheme > implementation. SRFI means Scheme Request for Implementation and not > Scheme Request for Copying, for a good reason, I think. > > PS: The module system of R6RS (whose semantics were mostly copied to > R7RS-small), which is the main entry point of running a Scheme program > in R[67]RS, has a limited expressiveness compared to what Racket (or > even R5RS) offers. The R[67]RS model does not allow to hook into the > primordial expander and evaluator (compare this with "#lang" in Racket > or with the ability to override the bindings of system identifiers in > R5RS), which makes it unclear how to turn the sample implementation of > this SRFI into a proper implementation. (Similarly, SRFI 9 can be > implemented in R5RS but not in R7RS (assuming it wasn't already) > because it has to redefine "vector?".) > > Maybe it is not too late for R7RS-large to improve this situation > considerably. This will also open the possibility to turn the module > language of R[67]RS, which is currently language outside Scheme, into > proper Scheme. > > > Am Mi., 8. Jan. 2025 um 17:01 Uhr schrieb Arthur A. Gleckler > <s...@speechcode.com>: > > > > On 8 Jan 2025, at 11:27, Shiro Kawai <shiro.ka...@gmail.com> wrote: > > > >> The SRFI needs to intercept the evaluation mechanism, so it's not > supposed to run as is in the bare REPL. > > > > > > On Wed, Jan 8, 2025 at 2:38 AM Daphne Preston-Kendal <d...@nonceword.org> > wrote: > > > >> > >> This should really be noted in the ‘Sample Implementation’ section of > the specification. > > > > > > Marc, would you be willing to craft some text to this effect (or review > text written by someone else)? I can add it to a post-finalization note. > > > > Thanks. >
