Dan,
I've been playing about with the new object stuff you added today, and
I've noticed a couple of problems.
Firstly, when your doing the initialization for a new ParrotClass PMC,
you create an Array to hold various other PMCs, but you don't size the
array; this means that when you later try to put things in it in
Parrot_new_class, it dies with:
Array index out of bounds!
The simplest solution is just to specify the size, as in the patch below.
------------
The other problem I noticed was that in a bunch of the ops you're using
code like:
PMC *class = VTABLE_get_pmc_keyed(interpreter,
interpreter->class_hash, key_new_string(interpreter, $2));
to get the ParrotClass PMC, and then seeing whether it's a valid class
by checking whether class is NULL. For instance, in findclass, you have:
if (VTABLE_get_pmc_keyed(interpreter, interpreter->class_hash,
key_new_string(interpreter, $2))) {
$1 = 1;
} else {
$1 = 0;
}
Unfortunately, this doesn't do what you intend. When you look up a
non-existent entry in a PerlHash, it creates and returns a PerlUndef PMC,
so the class pointer that you get is always non-NULL, even though it
isn't necessarily a ParrotClass. Consequently, in the example above,
findclass always returns true.
I'm not sure what the best fix is here.
Simon
--- classes/parrotclass.pmc Wed Jul 16 22:28:40 2003
+++ classes/parrotclass.pmc.new Wed Jul 16 22:28:35 2003
@@ -32,6 +32,8 @@
void init () {
/* Hang an array off the data pointer, empty of course */
PMC_data(SELF) = pmc_new(interpreter, enum_class_Array);
+ /* We will have five entries in this array */
+ VTABLE_set_integer_native(interpreter, (PMC*)PMC_data(SELF), (INTVAL)5);
/* No attributes to start with */
SELF->obj.u.int_val = 0;
/* But we are a class, really */
