wrap-expr/c: blame-source points to use site rather than definition

[lexi-lambda]

The blame-source field is supposed to describe where the contract is from, not where it’s used, even if the attachment of the contract is deferred. To illustrate, see the following behavior with contract-out:

#lang racket

(module a racket
  (provide (contract-out
            [f (-> integer? integer?)]))
  (define (f x)
    (add1 x)))

(module b racket
  (require (submod ".." a))
  (f 'not-an-integer))

(require 'b)

f: contract violation
  expected: integer?
  given: 'not-an-integer
  in: the 1st argument of
      (-> integer? integer?)
  contract from: (anonymous-module a)
  blaming: (anonymous-module b)
   (assuming the contract is correct)
  at: unsaved-editor:5.13

The equivalent example using expr/c puts the source at the use site, instead:

#lang racket

(module a racket
  (require syntax/parse/define)
  (provide f)
  (define-simple-macro (f e)
    #:declare e (expr/c #'integer?)
    (add1 e.c)))

(module b racket
  (require (submod ".." a))
  (f 'not-an-integer))

(require 'b)

f: contract violation
  expected: integer?
  given: 'not-an-integer
  in: macro argument contract
  contract from: (submod anonymous-module a)
  blaming: (submod (quote anonymous-module) b)
   (assuming the contract is correct)
  at: unsaved editor:12.5

The source location reported by wrap-expr/c is actually very useful, since normally that information isn’t reported anywhere, but it isn’t what blame-source is supposed to report.