Hildo Biersma writes: > If the number of objects that needs this is relatively small, we could > play a trick somewhat like the following (with small changes to the perl > compiler): > > 1. Break the filehandle object into two: a generic wrapper that uses > refcounting and forwards all calls, plus the actual filehandle object.
There seems to be a misconception about refcounting, and it's showing up here. People seem to think that you can do it to certain objects and not others. The fact is, right now when we're wiggling around PMCs we're just changing some pointers. But if there is one object that likes to be refcounted, then *every* pointer change needs to do the refcount dance. It's a big, potentially cache-blowing speed hit for the whole virtual machine. You can't do refcounting selectively. It's all or none. Of course, in C++, you can, because you tell C++ which pointers are refcountable pointers at compile time. If, by static analysis, we could tell Parrot when PMCs need refcounting before it assembled the bytecode, and also in some other places, we might be able to pull it off. But that's a halting problem problem if the language from which we're compiling is untyped. What I'd most like is to convince Larry to waive the timely destruction requirement. However, that doesn't really solve the problem for other languages that need timely destruction. Are there any? Luke
