Piers Cawley <[EMAIL PROTECTED]> wrote: > Well, I ported the following Scheme code to PIR. (The PIR is appended > to this message...
I've cleaned up the whole freelist handling code now and added your code as a test, which exposed one more bug: it didn't run with --gc-debug. That's fixed now too--if a return continuation morph's to a real continuation, it's context has to be marked. OTOH marking a return continuation context is wrong (after invokeing it, it's context is sooner or later outdated and might already point to freed objects). So it seems that we have to distinguish the two cases anyway, which is now done with one bit: recyling enabled or not. thanks for your continuation code, leo
