Conversation
| match-define match-define-values | ||
| match-letrec match-letrec-values | ||
| match/values match/derived match*/derived match-define-values |
There was a problem hiding this comment.
In the original code, match-define-values is provided twice.
| @@ -200,6 +200,8 @@ | |||
| (rename new-prop:procedure prop:procedure) | |||
| (rename #%app #%plain-app) | |||
| (rename lambda #%plain-lambda) | |||
| (rename λ #%plain-λ) | |||
There was a problem hiding this comment.
I'm not sure if it should be (rename λ #%plain-λ) or (rename lambda #%plain-λ).
| @@ -200,6 +200,8 @@ | |||
| (rename new-prop:procedure prop:procedure) | |||
| (rename #%app #%plain-app) | |||
| (rename lambda #%plain-lambda) | |||
| (rename λ #%plain-λ) | |||
| (rename case-lambda case-λ) | |||
There was a problem hiding this comment.
I'm not sure if case-λ should be defined in '#%kernel. I don't know how to add operators in '#%kernel, so I rename case-lambda to case-λ in pre-base.rkt.
There was a problem hiding this comment.
Changes to match-λ[*][*] is nice.
I really doubt that #%plain-λ is a good idea. No one will use #%plain-lambda anyway, since it's inferior than the regular lambda. It's only used inside the Racket internals, and also when people want to pattern match against expanded code. But in those two cases, #%plain-λ won't have much value.
My personal opinion is that this PR should be restricted to only changes in match-related form.
| @@ -331,8 +331,9 @@ | |||
| (append | |||
| (kernel-form-identifier-list) | |||
| (list | |||
| (quote-syntax #%app) ; racket/base app, as opposed to #%plain-app | |||
| (quote-syntax #%app) ; racket/base #%app, as opposed to #%plain-app | |||
There was a problem hiding this comment.
Please revert all changes to this file. free-identifier=? already takes renaming into account. There's no need to test against another renamed form.
There was a problem hiding this comment.
This file already handles similar processes for λ, as seen in: (or (free-identifier=? #'lam #'lambda) (free-identifier=? #'lam #'λ)). I assumed the relationship between lambda and λ was similar to that between case-lambda and case-λ, which is why I treated #%plain-λ and case-λ in the same manner. If supporting case-λ is considered, do we still need to revert all changes to this file?
There was a problem hiding this comment.
If case-λ is created via the renaming mechanism, then free-identifier=? will work fine on it. So yes, you should still revert all changes to this file.
Apparently λ is not created via the renaming mechanism from lambda, so that's why only one free-identifier=? would not suffice.
#lang racket
(module sub racket
(provide my-case-lambda
(rename-out [my-case-lambda my-case-λ]))
(define my-case-lambda 1))
(require syntax/parse/define
'sub)
(define-syntax-parse-rule (test a b)
#:do [(println (free-identifier=? #'a #'b))]
(void))
(test my-case-lambda my-case-λ) ;=> #t
(test lambda λ) ;=> #f
There was a problem hiding this comment.
I read the code, and if I understand correctly, lambda and λ in racket/base are respectively new-lambda and new-λ in kw.rkt:
(define-syntaxes (new-lambda new-λ)
(let ([new-lambda ...])
(values new-lambda new-lambda)))
Is it necessary to create a new PR and define new-λ through the renaming mechanism?
There was a problem hiding this comment.
For any big change like this (including this PR actually), I think it makes sense to get preliminary feedback from https://racket.discourse.group/ first. For example, it could be that people are somehow relying on lambda and λ not being free-identifier=? for some reasons, and the change would break their code.
But I agree that ideally lambda should be free-identifier=? to λ.
There was a problem hiding this comment.
It seems unlikely that changing how lambda and λ are bound is going to work but before a discussion, just checking to see if racket (and a bunch of packages, maybe starting with main-distribution) builds would be informative.
Initially, I added the λ forms based on search results for
|
|
This pull request has been mentioned on Racket Discourse. There might be relevant details there: https://racket.discourse.group/t/adding-more-unicode-forms-to-racket/2911/1 |
|
This pull request has been mentioned on Racket Discourse. There might be relevant details there: https://racket.discourse.group/t/renaming-mechanism-for-unicode-in-racket/2910/3 |
Checklist
Description of change
This PR introduces Unicode variants of common lambda forms in Racket. A previous PR #1301 suggested similar changes but was closed. This PR revives that idea.