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

Re: SRFI 97 and R6RS implementers?



Grant Rettke wrote:
In light of the modularity offered by R6RS, are all of the R6RS
implementers planning on collaborating on a single set of R6RS SRFI
implementations and reusing them?

I know of no such plans and personally doubt it, but I don't speak for any R6RS implementors.

Was SRFI 97 driven as part of such an effort?

Originally, this was one of the motivations for writing this SRFI. Several Scheme systems have minimal SRFI support and it is non-trivial (yet tedious and mostly uninteresting) to port all of the existing reference implementations. I wanted to provide high-quality implementations of all existing SRFIs so that R6RS-compatible Schemes could leverage a substantial library base with minimal effort.

I am beginning to doubt the utility of this, and I may remove this aspect from future drafts of the SRFI. It's not clear to me that R6RS-compatible implementations can always be high-quality. Also, some SRFIs admit several implementation strategies and it's not clear what the best one is. I am concerned that whatever implementation strategy is chosen, this will then be considered normative, which I certainly want to avoid. So I think maybe it is better to have R6RS-compatible systems provide their own implementations of past SRFIs. If the community would like to develop a portable collection of implementations, that is great and I'd be willing to contribute to it, but I have doubts that the SRFI process is how such a collection should be developed.

Of course, I am interested in others' perspectives on this.

David