The new intermediate representation uses an SSA representation for local variables.
Simple optimizations (such as dead code elimination, copy propagation and constant folding) have already been implemented.
The bytecode generation phase is not perfect yet (trivial register allocation is used which results in too many locals), but feature parity with the old implementation has been reached.
The new intermediate representation uses an SSA representation for local variables. Simple optimizations (such as dead code elimination, copy propagation and constant folding) have already been implemented.
The bytecode generation phase is not perfect yet (trivial register allocation is used which results in too many locals), but feature parity with the old implementation has been reached.