Revised^7 Report on Scheme, Large: Procedural Fascicle Draft is now public
r7rs.orgHello everyone.
Working Group 2 is pleased to announce the first draft of the second part of the R7RS-Large Foundations, the "Procedural Fascicle". This draft encompasses the familiar block programming forms, such as lambda, let, if, or, and set!.
The draft is available here:
https://r7rs.org/large/fascicles/proc/
The biggest new feature is the ability to mix definitions and expressions in bodies, such as the body of a lambda. For example, the following is now valid:
(define (map f lst)
(unless (list? lst)
(error 'map "not a list" lst))
(define (map* lst acc)
(if (null? lst)
(reverse acc)
(map* (cdr lst) (cons (f (car lst)) acc))))
(map* lst '()))
* Filing an issue on the R7RS-Large issue tracker <https://codeberg.org/scheme/r7rs> * Sending mail to the Working Group 2 mailing list <https://groups.google.com/g/scheme-reports-wg2> (you do not need a google account) * Sending mail to the Scheme Reports mailing list <https://scheme-reports.simplelists.com/> and <scheme-reports@scheme-reports.org> * Sending mail to the corresponding member Peter McGoron at <code@mcgoron.com>. I will forward your comment to the public issue tracker. Please indicate if you wish to be anonymous.
> The biggest new feature is the ability to mix definitions and expressions in bodies, such as the body of a lambda.
Looks like Rust. It helps macros a lot
I hope you don't take this question the wrong way: why is R7RS large taking so long? Is there a consensus problem or just a lack of resources? If it's the latter, is there any way a random scheme dabbler can help?
The committee needs to think long and hard about which features to propose in each docket and what those features should look like, a lot of discussion needs to take place in order to get each docket formed, then the wider public needs to get its chance to weigh in, and then a voting round needs to be held. That process may need to happen several times for a docket, and there's a bunch of dockets planned before R7RSL will be done.
Think of it this way: if the committee was more cavalier about this we'd have the familiar situation where we start with a great language and then every update worsens it until we end up with an abominable horror show.
Delimited continuation machinery? It seems like call/cc is becoming quite a bit more acknowledged as being overkill while the more restricted shift/reset, control/prompt, F-operator stuff, etc. compose better.
We plan to add delimited continuations in a later draft. What the actual API will look like is still in question. There is a proposal for a large API, like Racket's <https://srfi.schemers.org/srfi-226/> and a minimalistic API <https://srfi.schemers.org/srfi-248/> similar-ish to Guile's.