Export sexp_get_stack_trace

[dpapavas]

As far as I can see, there is no other way to get the current stack, at least not in a form that can be used arbitrarily instead of just printed to an output port. (There is sexp_exception_stack_trace of course, but I assume that's just for exceptions).

For concreteness, my use case is roughly this: I have procedures exported to Scheme, that can be called to create objects at the C/C++ level. I would like to inspect the stack at the time of creation, i.e. within the call to the exported procedure, in order to get the location in the source where the object is created, for use in diagnostic messages to the user. If there is some existing way of doing this, please let me know.

Since we're on the matter, I notice that stack traces (obtained via the function exported in this PR, or via sexp_exception_stack_trace) seem to be incomplete. Consider for instance a foreign procedure defined as

static sexp foo(sexp ctx, sexp self, sexp_sint_t n, sexp x)
{
    sexp_assert_type(ctx, sexp_realp, SEXP_NUMBER, x);

    /* ... */

    return SEXP_VOID;
}

and exported via

sexp_define_foreign(ctx, env, "foo", 1, foo);

If I call it from a file named, say test.scm, containing:

(foo "foo")

Then the stack trace I get from the thrown exception looks like:

#0 /usr/local/share/chibi/init-7.scm:1264
#1 /usr/local/share/chibi/init-7.scm:796

If I change the call to, say (foo 42), and get the stack trace via sexp_get_stack_trace from within the C foo function, I get the same trace. In any case, no stack frame with location within test.scm appears.

If I wrap foo in another function as in

(define (bar x) (foo x))

(display (bar "foo"))

I now get a more complete stack trace:

#0 test.scm:3
#1 /usr/local/share/chibi/init-7.scm:1264
#2 /usr/local/share/chibi/init-7.scm:796

But this seems to work only when the call is wrapped in another call (as in display here). Am I doing something wrong? If not, is there some way of getting a complete strack trace consistently? (If work is required on the Chibi-Scheme side, I could try to help out, given some pointers.)

[dpapavas]

Thanks for merging this. If you have any insight on the matter regaring stack trace completeness I mention above, please let me know. Feel free of course, to answer at your leisure; this is not particularly pressing. (Also, let me know if you'd prefer me to turn this into a separate issue.)