Aaron Sherman writes: > On Wed, 2005-04-27 at 14:38, Luke Palmer wrote: > > > There's still a lot of premature optimization going on [...] > > I'm surely guilty of one of them. I feel like the autothreading > > semantics of junctions will be way to expensive without the compiler > > knowing whether there a junction in a particular variable. > > Well, more to the point, autothreading of junctions will hit the wall of > Parrot duping the interpreter. That's probably not something you want to > suffer just to resolve a junction, is it?
What? Why will it do that? Luke