GitLab

Simon Peyton Jones pushed to branch wip/T26548 at Glasgow Haskell Compiler / GHC

Commits:

654d6ed9

by Simon Peyton Jones at 2025-12-24T13:33:08+00:00

Be a little less keen to inline

This commit changes interestingArg to treat lambda as NonTrivArg
rather than ValueArg.  That makes parser combinators a bit less
keen to inline.  E.g.

  <|> p1 p2 = \x -> case p1 x of
                      Yes -> ...
                      No  -> ...

If we have a call (<|> arg1 arg2) where arg1 is a parser, and hence
often visibly a lambda, it's no so great to inline <|>, because we
are still stuck on x.

This affects for example T17516

Just an experiment.

1 changed file:

compiler/GHC/Core/Opt/Simplify/Utils.hs

Changes:

compiler/GHC/Core/Opt/Simplify/Utils.hs

@@ -1016,7 +1016,7 @@ interestingArg env e = go env 0 e
      go env n (Lam v e)
         | isTyVar v             = go env n e
         | n>0                   = NonTrivArg     -- (\x.b) e   is NonTriv
 -       | otherwise             = ValueArg
 +       | otherwise             = NonTrivArg
      go _ _ (Case {})           = NonTrivArg
      go env n (Let b e)         = case go env' n e of
                                     ValueArg -> ValueArg
@@ -1028,7 +1028,7 @@ interestingArg env e = go env 0 e
         | isConLikeId v = ValueArg   -- Experimenting with 'conlike' rather that
                                      --    data constructors here
                                      -- DFuns are con-like; see Note [Conlike is interesting]
 -       | idArity v > n = ValueArg   -- Catches (eg) primops with arity but no unfolding
 +       | idArity v > n = NonTrivArg   -- Catches (eg) primops with arity but no unfolding
         | n > 0         = NonTrivArg -- Saturated or unknown call
         | otherwise  -- n==0, no value arguments; look for an interesting unfolding
         = case idUnfolding v of