[Larceny-users] weird error message compiling library

[Felix Klock]

Marco-

On Jun 29, 2009, at 4:41 PM, Marco Maggi wrote:

> http://github.com/marcomaggi/nausicaa/blob/25e23467cf948eb464704595e5a247d11bbd0158/scheme/src/libraries/random.sls



This snippet of code starting from line 343 of the above link seems  
pretty suspect to me; you have a unary definition of make-integer  
followed by an invocation with zero arguments.
       (define (make-integer n)
   (if (and (integer? n) (exact? n) (positive? n))
    (mod (next-integer) n)
    (assertion-violation 'integers-maker
    "range upper limit must be an exact positive integer" n)))

       (define (make-real)
   (* (inexact (make-integer)) const:1/2^32^2))

I used the error message emitted by Larceny as the hint to look here.

What do you think?

-Felix

p.s. Hypothesis: it could be that the other systems do not detect any  
problem here because the make-real procedure is never invoked in your  
code base.  I believe they are more conforming to the R6RS than  
Larceny is on this particular point; I also believe there is a way to  
coax Larceny into a mode where it will still compile the code and  
ignore the latent error hidden within.  If you want me to look up more  
information about this, let me know.