Re: LuaChip and GC

["Michael T. Richter" <mtr1966@...>]

> Does that imply that more cpu cycles would be needed than in current
> Lua GC implementation?  Remember, in portable devices that means
> battery life so I'd be personally inclined to the existing solution.

I haven't peered into the current GC implementation.  I doubt it would
require more cycles, however, since a trivial optimisation would have it
suspend the thread at the end of the treadmill and only reactivate it if
there's been a pointer write or a fresh allocation (the only points where a
new trace would be required at all).