[commit: ghc] master: Document missing dataToTag# . tagToEnum# rule (517c194)
git at git.haskell.org
git at git.haskell.org
Tue Feb 20 04:53:35 UTC 2018
Repository : ssh://git@git.haskell.org/ghc
On branch : master
Link : http://ghc.haskell.org/trac/ghc/changeset/517c194095064c123b18b779c96c0866e0b3a6d9/ghc
>---------------------------------------------------------------
commit 517c194095064c123b18b779c96c0866e0b3a6d9
Author: David Feuer <david.feuer at gmail.com>
Date: Mon Feb 19 23:48:22 2018 -0500
Document missing dataToTag# . tagToEnum# rule
Explain why we don't have a rule to optimize `dataToTag# (tagToEnum# x)`
to `x`.
[skip ci]
Reviewers: bgamari
Reviewed By: bgamari
Subscribers: rwbarton, thomie, carter
GHC Trac Issues: #14282
Differential Revision: https://phabricator.haskell.org/D4375
>---------------------------------------------------------------
517c194095064c123b18b779c96c0866e0b3a6d9
compiler/prelude/PrelRules.hs | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/compiler/prelude/PrelRules.hs b/compiler/prelude/PrelRules.hs
index c9a3bc7..73484b7 100644
--- a/compiler/prelude/PrelRules.hs
+++ b/compiler/prelude/PrelRules.hs
@@ -909,6 +909,19 @@ dataToTagRule = a `mplus` b
guard $ ty1 `eqType` ty2
return tag
+ -- Why don't we simplify tagToEnum# (dataToTag# x) to x? We would
+ -- like to, but it seems tricky. See #14282. The trouble is that
+ -- we never actually see tagToEnum# (dataToTag# x). Because dataToTag#
+ -- is can_fail, this expression is immediately transformed into
+ --
+ -- case dataToTag# @T x of wild
+ -- { __DEFAULT -> tagToEnum# @T wild }
+ --
+ -- and wild has no unfolding. Simon Peyton Jones speculates one way around
+ -- might be to arrange to give unfoldings to case binders of CONLIKE
+ -- applications and mark dataToTag# CONLIKE, but he doubts it's really
+ -- worth the trouble.
+
-- dataToTag (K e1 e2) ==> tag-of K
-- This also works (via exprIsConApp_maybe) for
-- dataToTag x
More information about the ghc-commits
mailing list