enable-warnings time-travelling shenanigans

[Idiomdrottning]

(import miscmacros (chicken condition))
(define-syntax-rule

(bound-procedure? name)
(handle-exceptions

ex

#f
(procedure? name)))

(enable-warnings #f)
(define wtf (bound-procedure? if))
(enable-warnings #t)
wtf

;; ↑↑ quietly and politely evals to #f

(begin

(enable-warnings #f)
(define wtf (bound-procedure? if))

(enable-warnings #t)
wtf)

;; ↑↑ expands to a let form (which is fine of course) and complains in
;; the REPL because warnings apply retroactively to the entire form.

I've tried enabling the warnings via (let ((ret ...)) enable ret) dynamic-wind or via force/delay or via just a lambda that gets called, but no matter how I do it, it retroactively wants to print warnings.

I even did

(define (re-enable-warnings val ret)

(enable-warnings val)
ret)

(re-enable-warnings #f (bound-procedure? if))

;;; politely evals to #f

(re-enable-warnings #t (bound-procedure? if))

;;; causes a ruckus

The time travelling super powers of call-with-current-continuation that the REPL uses is haunting me…!

This just can't be right!

[Idiomdrottning]

Oh, ignore-errors already exist so forget my custom bound-procedure.

Here is an already miscmacros-established way to get the same weird behavior:

(import miscmacros)

(enable-warnings #f)
(define wtf (ignore-errors (procedure? if)))
(enable-warnings #t)
wtf

;; politely evals #f

(begin

(enable-warnings #f)
(define wtf (ignore-errors (procedure? if)))
(enable-warnings #t)
wtf)

;; notices at the REPL

[felix winkelmann]

The "begin" for is expanded before the enable-warnings parameter is changed, this is simply because expansion time precedes run-time. You need to define syntax here that performs the parameter-change at expansion time.