I just committed a change to the code that moves the registers into the interpreter structure rather than accessing the things indirectly as we were before. Pushes and pops are a little slower as there are memcpys involved now, but register access itself should be faster as there's one fewer level of indirection.
Yes, four of the stacks.t tests fail. We need to patch that one up--I'll be taking a shot in the morning when things have had a chance to settle a bit, but I wanted to get this bit in to start. -- Dan --------------------------------------"it's like this"------------------- Dan Sugalski even samurai [EMAIL PROTECTED] have teddy bears and even teddy bears get drunk