Expander tries to take the car of some #{thingamabob ...}

[Gwen Weinholt]

I want to learn more about how syntax-case etc works, so I'm reading Waddell and Dybvig's paper "Extending the Scope of Syntactic Abstraction". There's an example in section 5.2 that I thought wouldn't work in R6RS, so I modified it a little and here's what happened:

Ikarus Scheme version 0.0.4-rc1+, 64-bit (revision 1834, build 2009-08-01)
Copyright (c) 2006-2009 Abdulaziz Ghuloum

> (library (M)
  (export a b)
  (import (rnrs))
  (define-syntax a
    (syntax-rules ()
      [(_ e0 e1 ...) (e0 (quote c) e1 ... )]))
  (define b (lambda () c))
  (define c 3)
  (lambda () (set! c 5))) ;make Ikarus believe c is mutable
> (import (M))
> (let ()
  (define-syntax f
    (syntax-rules ()
      [(_ (any id)) id]
      [(_ (any id) value) (set! id value)]))
  (let ([original (a f)])
    (a f 4)
    (list original (b)))) ; (3 4)
Unhandled exception
 Condition components:
   1. &assertion
   2. &who: car
   3. &message: "argument does not have required pair structure"
   4. &irritants: (#{c |&qTay!6SOGg0MwBL|})
>

I'm still pretty sure this isn't supposed to work, but the exception raised indicates a bug in the expander, rather than in the code above.

Btw, it works without error in Ypsilon. :)

[Abdulaziz Ghuloum (aghuloum)]

On Aug 1, 2009, at 8:41 PM, Göran Weinholt wrote:

> Btw, it works without error in Ypsilon. :)

I'm not sure what you mean by "it works", but I fixed the crash in
revision 1835.

Aziz,,,

** Affects: ikarus
    Status: Fix Committed

[Gwen Weinholt (weinholt)]

> > Btw, it works without error in Ypsilon. :)

> I'm not sure what you mean by "it works", but I fixed the crash in
> revision 1835.

I tried the example from the bug report in Ypsilon 0.9.6-trunk/r503 and it returned (3 4).

Thanks for the fix, it's clear to me now that Ypsilon is in error.