From: simonpj@microsoft.com <unknown>
Date: Wed, 6 Sep 2006 11:52:42 +0000 (+0000)
Subject: Make record selection look cheap
X-Git-Tag: Before_FC_branch_merge~62
X-Git-Url: http://git.megacz.com/?p=ghc-hetmet.git;a=commitdiff_plain;h=061d33c9923500d169570171947335053b63755f

Make record selection look cheap

exprIsCheap was already making record selection look cheap, esp for
dictionary selection; e.g.
   (==) dEq

The idea is to allow those record selections to migrate inside a
lambda, because on the whole it's more expensive to share than
repeat them.

But sometimes dictionary selection is nested; e.g.
   (==) ($p1Num dNum)

This turned out to make a (fragile) rewrite rule fire a little
more robustly, but it seems like a good idea in any case.

This patch makes nested record selections also look cheap.
---

diff --git a/compiler/coreSyn/CoreUtils.lhs b/compiler/coreSyn/CoreUtils.lhs
index 69f78fa..c59c751 100644
--- a/compiler/coreSyn/CoreUtils.lhs
+++ b/compiler/coreSyn/CoreUtils.lhs
@@ -476,7 +476,7 @@ exprIsCheap other_expr 	-- Applications and variables
  	-- to bother to check the number of args
  
     --------------
-    go_sel [arg] = exprIsTrivial arg	-- I'm experimenting with making record selection
+    go_sel [arg] = exprIsCheap arg	-- I'm experimenting with making record selection
     go_sel other = False 	 	-- look cheap, so we will substitute it inside a
  					-- lambda.  Particularly for dictionary field selection.
   		-- BUT: Take care with (sel d x)!  The (sel d) might be cheap, but