Change isUnliftedTyCon to marshalablePrimTyCon (#20401)

isUnliftedTyCon was used in three places: Ticky, Template Haskell
and FFI checks.

It was straightforward to remove it from Ticky and Template Haskell.

It is now used in FFI only and renamed to marshalablePrimTyCon.

Previously, it was fetching information from a field
in PrimTyCon called is_unlifted. Instead, I've changed the code
to compute liftedness based on the kind.

isFFITy and legalFFITyCon are removed. They were only referred from
an old comment that I removed.

There were three functions to define a PrimTyCon, but the only difference
was that they were setting is_unlifted to True or False.
Everything is now done in mkPrimTyCon.

I also added missing integer types in Ticky.hs, I think it was an oversight.

Fixes #20401

1 files changed, 0 insertions, 103 deletions

diff --git a/compiler/GHC/Iface/Tidy.hs b/compiler/GHC/Iface/Tidy.hs
index 41b1ad6b9e..ed5e99805f 100644
--- a/compiler/GHC/Iface/Tidy.hs
+++ b/compiler/GHC/Iface/Tidy.hs
@@ -1338,106 +1338,3 @@ tidyTopIdInfo uf_opts rhs_tidy_env name orig_rhs tidy_rhs idinfo show_unfold
     -- it to the top level. So it seems more robust just to
     -- fix it here.
     arity = exprArity orig_rhs
-
-{-
-************************************************************************
-*                                                                      *
-                  Old, dead, type-trimming code
-*                                                                      *
-************************************************************************
-
-We used to try to "trim off" the constructors of data types that are
-not exported, to reduce the size of interface files, at least without
--O.  But that is not always possible: see the old Note [When we can't
-trim types] below for exceptions.
-
-Then (#7445) I realised that the TH problem arises for any data type
-that we have deriving( Data ), because we can invoke
-   Language.Haskell.TH.Quote.dataToExpQ
-to get a TH Exp representation of a value built from that data type.
-You don't even need {-# LANGUAGE TemplateHaskell #-}.
-
-At this point I give up. The pain of trimming constructors just
-doesn't seem worth the gain.  So I've dumped all the code, and am just
-leaving it here at the end of the module in case something like this
-is ever resurrected.
-
-
-Note [When we can't trim types]
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-The basic idea of type trimming is to export algebraic data types
-abstractly (without their data constructors) when compiling without
--O, unless of course they are explicitly exported by the user.
-
-We always export synonyms, because they can be mentioned in the type
-of an exported Id.  We could do a full dependency analysis starting
-from the explicit exports, but that's quite painful, and not done for
-now.
-
-But there are some times we can't do that, indicated by the 'no_trim_types' flag.
-
-First, Template Haskell.  Consider (#2386) this
-        module M(T, makeOne) where
-          data T = Yay String
-          makeOne = [| Yay "Yep" |]
-Notice that T is exported abstractly, but makeOne effectively exports it too!
-A module that splices in $(makeOne) will then look for a declaration of Yay,
-so it'd better be there.  Hence, brutally but simply, we switch off type
-constructor trimming if TH is enabled in this module.
-
-Second, data kinds.  Consider (#5912)
-     {-# LANGUAGE DataKinds #-}
-     module M() where
-     data UnaryTypeC a = UnaryDataC a
-     type Bug = 'UnaryDataC
-We always export synonyms, so Bug is exposed, and that means that
-UnaryTypeC must be too, even though it's not explicitly exported.  In
-effect, DataKinds means that we'd need to do a full dependency analysis
-to see what data constructors are mentioned.  But we don't do that yet.
-
-In these two cases we just switch off type trimming altogether.
-
-mustExposeTyCon :: Bool         -- Type-trimming flag
-                -> NameSet      -- Exports
-                -> TyCon        -- The tycon
-                -> Bool         -- Can its rep be hidden?
--- We are compiling without -O, and thus trying to write as little as
--- possible into the interface file.  But we must expose the details of
--- any data types whose constructors or fields are exported
-mustExposeTyCon no_trim_types exports tc
-  | no_trim_types               -- See Note [When we can't trim types]
-  = True
-
-  | not (isAlgTyCon tc)         -- Always expose synonyms (otherwise we'd have to
-                                -- figure out whether it was mentioned in the type
-                                -- of any other exported thing)
-  = True
-
-  | isEnumerationTyCon tc       -- For an enumeration, exposing the constructors
-  = True                        -- won't lead to the need for further exposure
-
-  | isFamilyTyCon tc            -- Open type family
-  = True
-
-  -- Below here we just have data/newtype decls or family instances
-
-  | null data_cons              -- Ditto if there are no data constructors
-  = True                        -- (NB: empty data types do not count as enumerations
-                                -- see Note [Enumeration types] in GHC.Core.TyCon
-
-  | any exported_con data_cons  -- Expose rep if any datacon or field is exported
-  = True
-
-  | isNewTyCon tc && isFFITy (snd (newTyConRhs tc))
-  = True   -- Expose the rep for newtypes if the rep is an FFI type.
-           -- For a very annoying reason.  'Foreign import' is meant to
-           -- be able to look through newtypes transparently, but it
-           -- can only do that if it can "see" the newtype representation
-
-  | otherwise
-  = False
-  where
-    data_cons = tyConDataCons tc
-    exported_con con = any (`elemNameSet` exports)
-                           (dataConName con : dataConFieldLabels con)
--}