At 09:16 PM 7/17/2002 -0400, Dan Sugalski wrote: >Melvin (and the rest of you... :) > >Are subroutine PMCs at a point where we can use them? If not, that's fine, >but if so, they can go on the list 'o things. (And we can add global >variables, if they didn't make it in the 0.0.6 release notes)
Subs, co-routines and continuations are at a very limited, but functional state. Basically you can create a PMC with a bytecode address or label and use the call/callco/callcc ops on it. We can: -capture and restore continuations. -call, yield and re-call coroutines. -call plain subs Problems: -The ops are wrong, they should be PMC methods -For continuations the stack frames currently won't get collected (I started work on it). Also you have to turn off the garbage collector to play around with continuations, since I haven't put in the custom mark routine yet. -Since we have no symbol tables yet, you have to create them at runtime. -We have no lexicals so only registers and stack work for the closure variables -The save/restore of a context is currently a bit more expensive than it has to be, but it works. -COW copies the whole stack for continuations, but I think I already put in a patch that reworked stacks enough to allow COW on a chunk by chunk basis. -About 100 other things Sometime in the last couple of months I mailed samples of co-routines and continuations to the list. I'll try to dig that up if you don't have it. -Melvin
