From: Jiri S. <jir...@bl...> - 2008-07-09 20:33:51
|
Albert Graef wrote: > Yeah, that's weird. Entering the same expression again it executes > instantly. And if I compare the LLVM IR of set (1..2) and set (1..10), > it's exactly the same assembler code except for the second int constant, > so the JIT would basically generate the same native machine code in both > cases. Frankly, I have no idea what's going on there. :) > At least we should remember that complicated guards are even much bigger disaster than complicated pattern matches... Jiri |