On Fri, Sep 24, 2010 at 12:58 PM, Simon Peyton-Jones
Can you give some examples to back up this claim? I am skeptical
I can look into it again after ICFP. I did spend some time trying to tease apart the performance improvements we got from inlining foldl'. From what I remember, the biggest gain from inlining is that the higher-order argument is known, but there's some additional gain from unboxing the accumulator (and some additional gain from inlining simple higher-order arguments like (+) into the body of foldl').
That’s what the auto-specialisation I described does.
You can’t write a SECIALISE pragma in the libray for a type that is defined by the client. This is very very common
Right, I think that's the reason we rely on INLINING pragmas so much. Inlining gives us the effect of specialize, but at types not known to the library author. I don't think we really want to force inlining, but want to say something like this: foldl' :: (a -> b -> a) -> a -> [b] -> a foldl' = ... {-# SPECIALIZABLE foldl' #-} and then have foldl' be specialized in the client module at whatever type it's called at, but only once for each type (instead of once per call site). -- Johan