From: simonpj@microsoft.com <unknown>
Date: Wed, 17 Nov 2010 10:05:10 +0000 (+0000)
Subject: Make (^) and (^^) INLINABLE
X-Git-Url: http://git.megacz.com/?p=ghc-base.git;a=commitdiff_plain;h=f2e36773ae93fa526cc05981e8f928f14bf612c8

Make (^) and (^^) INLINABLE

This makes them perform well robustly
(e.g. in test perf/should_run/MethSharing)
rather than relying on a rather delicate
let-floating.  See Note [Inlining (^) in Real.lhs
---

diff --git a/GHC/Real.lhs b/GHC/Real.lhs
index b3316d1..68e3784 100644
--- a/GHC/Real.lhs
+++ b/GHC/Real.lhs
@@ -424,6 +424,7 @@ odd             =  not . even
         Integer -> Integer -> Integer,
         Integer -> Int -> Integer,
         Int -> Int -> Int #-}
+{-# INLINABLE (^) #-}    -- See Note [Inlining (^)]
 (^) :: (Num a, Integral b) => a -> b -> a
 x0 ^ y0 | y0 < 0    = error "Negative exponent"
         | y0 == 0   = 1
@@ -439,8 +440,20 @@ x0 ^ y0 | y0 < 0    = error "Negative exponent"
 
 -- | raise a number to an integral power
 (^^)            :: (Fractional a, Integral b) => a -> b -> a
+{-# INLINABLE (^^) #-}         -- See Note [Inlining (^)
 x ^^ n          =  if n >= 0 then x^n else recip (x^(negate n))
 
+{- Note [Inlining (^)
+   ~~~~~~~~~~~~~~~~~~~~~
+   The INLINABLE pragma allows (^) to be specialised at its call sites.
+   If it is called repeatedly at the same type, that can make a huge
+   difference, because of those constants which can be repeatedly
+   calculated.
+
+   Currently the fromInteger calls are not floated because we get
+             \d1 d2 x y -> blah
+   after the gentle round of simplification. -}
+
 -------------------------------------------------------
 -- Special power functions for Rational
 --