Re: [Chicken-users] error message question

[Peter Keller]

On Mon, Aug 12, 2002 at 09:53:06AM +0200, felix wrote:
> > ;; dereference an mp_exp_t_p type into an mp_exp_t
> > (define mp_exp_t_p->mp_exp_t
> >     (define-foreign-lambda* mp_exp_t ((mp_exp_t_p val))
> >             "return((mp_exp_t)*val)"))
> 
> What's `define-foreign-lambda*'?

AHA! That's the bug! I meant to say 'foreign-lambda*' there. Too sleepy...
I suppose I didn't catch it because things like "define-foreign-type" are
in the API and it looked similar enough to be plausible. I guess I can
sorta see why the compiler did what it did, but it should check the validity
of the function BEFORE the validity of its evaluated arguments so it could have
told me the real error, which was define-foreign-lambda* didn't exist.

> Why there is no stacktrace produced, I don't know. Have you compiled
> with `-no-trace'?

Not explicitly. I compiled this revision from the head of the cvs trunk
and made it so "-Wall -g -Wno-unused" is returned by chicken-config
-cflags.

In other news, I am well on my way to getting a gmp_ffi 0.0.3 out! I've just
added support for some of the harder API calls and wrote the validation
code for them. I figure a week or two (finding time is hard in recent weeks)
before I've done 99% of the normal MP API and written the validation code for
all of it.

I suspect with the knowledge I picked up in supporting:
char* mpf_get_str(char *str, mp_exp_t *exp, int base, size_t n_digits, mpf_t op)

I can do the rest of the high-level API without too much trouble, I think.

My next project, I think, is zlib.

-pete