[racket] does anyone have an example of a closure cycle in zo-parse?

From: Matthew Flatt (mflatt at cs.utah.edu)
Date: Sun May 8 20:39:12 EDT 2011

Do you mean the same `code' field but a different `gen-id' field? I
think that will never happen.

The `gen-id' field just makes the `eq?'-level identity of constant
closures more apparent (that is, two closures should have the same
`gen-id' if an only if they are `eq?'). At the same time, a `closure'
wrapper is generated by the compiler as soon as it sees a closed
`lambda' form, so the form wouldn't be referenced in multiple places
except through the `closure' wrapper.

At Sun, 8 May 2011 18:50:06 -0400, Danny Yoo wrote:
> In Racket 5.1.1's compiler/zo-parse module, I see that the
> documentation on the 'closure' structure says that there can be cycles
> due to recursive closure calling.  I can see this, but I haven't been
> able to figure out why there would be different instances of closures
> with different gen-ids.  Does anyone have an example of one?
> Here's the code I'm using to play with 5.1.1's compiler/zo-parse module.
> ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
> ;;;;;;;;;;;;
> #lang racket/base
> (require compiler/zo-parse
>             (for-syntax racket/base))
> (define (run-zo-parse stx)
>   (parameterize ([current-namespace (make-base-namespace)])
>     (let ([bc (compile stx)]
>           [op (open-output-bytes)])
>       (write bc op)
>       (zo-parse (open-input-bytes (get-output-bytes op))))))
> ;; A cyclic structure should show here:
> (run-zo-parse #'(letrec ([g (lambda () (g))]) (g)))
> ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
> ;;;;;;;;;;;;;;;
> _________________________________________________
>   For list-related administrative tasks:
>   http://lists.racket-lang.org/listinfo/users

Posted on the users mailing list.