[Git][ghc/ghc][wip/T22948] Don't generate datacon wrappers for `type data` declarations
Ryan Scott (@RyanGlScott)
gitlab at gitlab.haskell.org
Sat Feb 11 02:11:11 UTC 2023
Ryan Scott pushed to branch wip/T22948 at Glasgow Haskell Compiler / GHC
Commits:
28b499e5 by Ryan Scott at 2023-02-10T21:10:36-05:00
Don't generate datacon wrappers for `type data` declarations
Data constructor wrappers only make sense for _value_-level data constructors,
but data constructors for `type data` declarations only exist at the _type_
level. This patch does the following:
* The criteria in `GHC.Types.Id.Make.mkDataConRep` for whether a data
constructor receives a wrapper now consider whether or not its parent data
type was declared with `type data`, omitting a wrapper if this is the case.
* Now that `type data` data constructors no longer receive wrappers, there is a
spot of code in `refineDefaultAlt` that panics when it encounters a value
headed by a `type data` type constructor. I've fixed this with a special case
in `refineDefaultAlt` and expanded `Note [Refine DEFAULT case alternatives]`
to explain why we do this.
Fixes #22948.
- - - - -
7 changed files:
- compiler/GHC/Core/Utils.hs
- compiler/GHC/Rename/Module.hs
- compiler/GHC/Types/Id/Make.hs
- + testsuite/tests/type-data/should_compile/T22948b.hs
- testsuite/tests/type-data/should_compile/all.T
- + testsuite/tests/type-data/should_run/T22948a.hs
- testsuite/tests/type-data/should_run/all.T
Changes:
=====================================
compiler/GHC/Core/Utils.hs
=====================================
@@ -728,9 +728,8 @@ refineDefaultAlt :: [Unique] -- ^ Uniques for constructing new binders
refineDefaultAlt us mult tycon tys imposs_deflt_cons all_alts
| Alt DEFAULT _ rhs : rest_alts <- all_alts
, isAlgTyCon tycon -- It's a data type, tuple, or unboxed tuples.
- , not (isNewTyCon tycon) -- We can have a newtype, if we are just doing an eval:
- -- case x of { DEFAULT -> e }
- -- and we don't want to fill in a default for them!
+ , not (isNewTyCon tycon) -- Exception 1 in Note [Refine DEFAULT case alternatives]
+ , not (isTypeDataTyCon tycon) -- Exception 2 in Note [Refine DEFAULT case alternatives]
, Just all_cons <- tyConDataCons_maybe tycon
, let imposs_data_cons = mkUniqSet [con | DataAlt con <- imposs_deflt_cons]
-- We now know it's a data type, so we can use
@@ -815,6 +814,39 @@ with a specific constructor is desirable.
`imposs_deflt_cons` argument is populated with constructors which
are matched elsewhere.
+There are two exceptions where we avoid refining a DEFAULT case:
+
+* Exception 1: Newtypes
+
+ We can have a newtype, if we are just doing an eval:
+
+ case x of { DEFAULT -> e }
+
+ And we don't want to fill in a default for them!
+
+* Exception 2: `type data` declarations
+
+ The data constructors for a `type data` declaration (see
+ Note [Type data declarations] in GHC.Rename.Module) do not exist at the
+ value level. Nevertheless, it is possible to strictly evaluate a value
+ whose type is a `type data` declaration. Test case
+ type-data/should_compile/T2294b.hs contains an example:
+
+ type data T a where
+ A :: T Int
+
+ f :: T a -> ()
+ f !x = ()
+
+ We want to generate the following Core for f:
+
+ f = \(@a) (x :: T a) ->
+ case x of
+ __DEFAULT -> ()
+
+ Namely, we do _not_ want to match on `A`, as it doesn't exist at the value
+ level!
+
Note [Combine identical alternatives]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
If several alternatives are identical, merge them into a single
=====================================
compiler/GHC/Rename/Module.hs
=====================================
@@ -2132,6 +2132,20 @@ The main parts of the implementation are:
`type data` declarations. When these are converted back to Hs types
in a splice, the constructors are placed in the TcCls namespace.
+* A `type data` declaration _never_ generates wrappers for its data
+ constructors, as they only make sense for value-level data constructors.
+ This extends to `type data` declarations implemented as GADTs, such as
+ this example from #22948:
+
+ type data T a where
+ A :: T Int
+ B :: T a
+
+ If `T` were an ordinary `data` declaration, then `A` would have a wrapper
+ to account for the GADT-like equality in its return type. Because `T` is
+ declared as a `type data` declaration, however, the wrapper is omitted.
+ See `wrapped_reqd` in GHC.Types.Id.Make.mkDataConRep` for the place where
+ this check is implemented.
-}
warnNoDerivStrat :: Maybe (LDerivStrategy GhcRn)
=====================================
compiler/GHC/Types/Id/Make.hs
=====================================
@@ -789,6 +789,10 @@ mkDataConRep dc_bang_opts fam_envs wrap_name data_con
(unboxers, boxers) = unzip wrappers
(rep_tys, rep_strs) = unzip (concat rep_tys_w_strs)
+ -- This is True if the data constructor or class dictionary constructor
+ -- needs a wrapper. This wrapper is injected into the program later in the
+ -- CoreTidy pass. See Note [Injecting implicit bindings] in GHC.Iface.Tidy,
+ -- along with the accompanying implementation in getTyConImplicitBinds.
wrapper_reqd =
(not new_tycon
-- (Most) newtypes have only a worker, with the exception
@@ -796,13 +800,27 @@ mkDataConRep dc_bang_opts fam_envs wrap_name data_con
&& (any isBanged (ev_ibangs ++ arg_ibangs)))
-- Some forcing/unboxing (includes eq_spec)
|| isFamInstTyCon tycon -- Cast result
- || dataConUserTyVarsNeedWrapper data_con
+ || (dataConUserTyVarsNeedWrapper data_con
-- If the data type was written with GADT syntax and
-- orders the type variables differently from what the
-- worker expects, it needs a data con wrapper to reorder
-- the type variables.
- -- See Note [Data con wrappers and GADT syntax].
- -- NB: All GADTs return true from this function
+ --
+ -- NB: All GADTs return true from this function, but there
+ -- is one exception that we must check below.
+ && not (isTypeDataTyCon tycon))
+ -- An exception to this rule is `type data` declarations.
+ -- Their data constructors only live at the type level and
+ -- therefore do not need wrappers.
+ -- See Note [Type data declarations] in GHC.Rename.Module.
+ --
+ -- Note that the other checks in this definition will
+ -- return False for `type data` declarations, as:
+ --
+ -- - They cannot be newtypes
+ -- - They cannot have strict fields
+ -- - They cannot be data family instances
+ -- - They cannot have datatype contexts
|| not (null stupid_theta)
-- If the data constructor has a datatype context,
-- we need a wrapper in order to drop the stupid arguments.
=====================================
testsuite/tests/type-data/should_compile/T22948b.hs
=====================================
@@ -0,0 +1,8 @@
+{-# LANGUAGE TypeData #-}
+module T22948b where
+
+type data T a where
+ A :: T Int
+
+f :: T a -> ()
+f !x = ()
=====================================
testsuite/tests/type-data/should_compile/all.T
=====================================
@@ -5,3 +5,4 @@ test('TDGoodConsConstraints', normal, compile, [''])
test('TDVector', normal, compile, [''])
test('TD_TH_splice', js_broken(22576), compile, [''])
test('T22315a', [extra_files(['T22315a/'])], multimod_compile, ['T22315a.Lib T22315a.Main', '-v0'])
+test('T22948b', normal, compile, [''])
=====================================
testsuite/tests/type-data/should_run/T22948a.hs
=====================================
@@ -0,0 +1,8 @@
+{-# LANGUAGE TypeData #-}
+module Main where
+
+type data T a where
+ A :: T Int
+ B :: T a
+
+main = return ()
=====================================
testsuite/tests/type-data/should_run/all.T
=====================================
@@ -1,3 +1,4 @@
test('T22332a', exit_code(1), compile_and_run, [''])
test('T22315b', extra_files(['T22315b.hs']), ghci_script, ['T22315b.script'])
test('T22500', normal, compile_and_run, [''])
+test('T22948a', normal, compile_and_run, [''])
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/28b499e59c6341e82644a8a6975a88c727141aef
--
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/commit/28b499e59c6341e82644a8a6975a88c727141aef
You're receiving this email because of your account on gitlab.haskell.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-commits/attachments/20230210/0daed883/attachment-0001.html>
More information about the ghc-commits
mailing list