freenode/#clasp - IRC Chatlog
Search
12:16:35
drmeister
Macintosh HD 1 is the hard drive of my old computer accessed through the cable. The numbers are slow and all over the place.
12:55:56
Bike
okay so here's what i'm thinking about inlining. first off, if we have the function body and the call site, there's no way we "can't" inline. We just need to make sure we allocate the ex-callee's environment at the ex-call site - conceptually, inline it.
12:56:19
Bike
this might not work for partial inlining, but i'm not really concerned with partial inlining
12:57:32
Bike
second, we've been talking about unifying handling of simple variables with closure variables, but i think it would make more sense to unify closure variables with data structures. The only difference between a cell and a cons is that one can be enclosed over and the other can be used as an argument. I don't think any analysis that could cover one couldn't cover the other.