Re: [GHC] #13016: SPECIALIZE INLINE doesn't necessarily inline specializations of a recursive function

17 Jan 2018

      #13016: SPECIALIZE INLINE doesn't necessarily inline specializations of a recursive
function
-------------------------------------+-------------------------------------
        Reporter:  nfrisby           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.0.1
      Resolution:                    |             Keywords:  Inlining
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Runtime           |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #13014            |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by andrewthad):

 I also feel the behavior of `SPECIALIZE INLINE` is weird in this
 situation, and I would prefer that it not be marked as a loop breaker.
 Here's an example that I was toying with that led me to this ticket:

 {{{!#hs
 {-# language DataKinds #-}
 {-# language GADTs #-}
 {-# language KindSignatures #-}

 {-# OPTIONS_GHC -O2 -fforce-recomp -ddump-simpl -dsuppress-all #-}
 import Data.Kind

 data Nat = Succ Nat | Zero

 data SNat :: Nat -> Type where
   SZero :: SNat 'Zero
   SSucc :: SNat n -> SNat ('Succ n)

 {-# SPECIALISE INLINE exponentiate :: SNat ('Succ n) -> Int -> Int #-}
 {-# SPECIALISE INLINE exponentiate :: SNat 'Zero -> Int -> Int #-}
 exponentiate :: SNat n -> Int -> Int
 exponentiate SZero x = 1
 exponentiate (SSucc s) x = x * (exponentiate s x)

 main :: IO ()
 main = print (exponentiate (SSucc (SSucc (SSucc (SSucc SZero)))) 3)
 }}}

 I would expect that the call to `exponentiate` be supercompiled, but it is
 not.

-- 
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/13016#comment:11
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler