Declare deprecated aliases

[jackfirth]

This will allow Resyntax to replace uses of these legacy forms with their modern names.

[github-actions]

Resyntax analyzed 4 files in this pull request and has added suggestions.

[github-actions]

arrow-contract-with-rest-to-arrow-contract-with-ellipses: This ->* contract can be rewritten using -> with ellipses.

Suggested change

	[filebox (-> (or/c core:element? string?) pre-flow? ... block?)]))

Debugging details

Textual replacement

(line-replacement
  #:new-lines
    '#("          [filebox (-> (or/c core:element? string?) pre-flow? ... block?)]))")
  #:original-lines
    '#("          [filebox (((or/c core:element? string?)) () #:rest (listof pre-flow?) . ->* . block?)]))")
  #:start-line 27)

Syntactic replacement

(syntax-replacement
  #:introduction-scope #<procedure:...and/syntax-local.rkt:148:2>
  #:new-syntax
    #<syntax:/home/runner/.local/share/racket/8.17.0.6/pkgs/resyntax/default-recommendations/contract-shortcuts.rkt:58:2 (-> (or/c core:element? string?) pre-flow? ... block?)>
  #:original-syntax
    #<syntax:scribble-lib/scribble/private/manual-style.rkt:27:19 (->* ((or/c core:element? string?)) () #:rest (listof pre-flow?) block?)>
  #:source
    (file-source
     #<path:/home/runner/work/scribble/scribble/scribble-lib/scribble/private/manual-style.rkt>))

[mflatt]

I think this needs to be updated for a change to define-deprecated-alias, which now requires that racketblock is defined before schemeblock is defined as an alias, for example.

[jackfirth]

Hmm. Is there a way for define-deprecated-alias to raise an error if the target is unbound in the surrounding module or definition context and not require the target be bound before the define-deprecated-alias form? That would be the friendliest API for users.

[mflatt]

You could use syntax-local-lift-module-end-declaration when expanding within a module (i.e., when syntax-transforming-module-expression? returns true).

[jackfirth]

Is there any way to achieve the same thing in an internal definition context?

[mflatt]

Do you mean an internal definition context that's in a module? Using syntax-local-lift-module-end-declaration should do the right thing when lifting a reference that happens to be in a definition context, as long as that definition context is also in a module as illustrated below. But maybe I don't understand what you're after.

#lang racket

(define-syntax (check-later stx)
  (syntax-case stx ()
    [(_ id)
     (when (syntax-transforming-module-expression?)
       (syntax-local-lift-module-end-declaration #'(check-now id)))
     #'(void)]))

(define-syntax (check-now stx)
  (syntax-case stx ()
    [(_ id)
     (unless (identifier-binding #'id)
       (raise-syntax-error #f "not good" #'id))
     #'(void)]))

(let ([ok "ok"])
  (check-later ok))

(let ()
  (check-later ok)
  (define ok "ok")
  (void))

(let ()
  (define something-else 0)
  (check-later ok-too))

(define ok-too 10)

[jackfirth]

I drafted racket/racket#5289 to clarify what I'd like to work. I think syntax-local-lift-module-end-declaration will work correctly for the test cases I wrote, but I haven't tried it yet.