GitLab

The UserTypeCtxt data type defines a context for TH brackets:

data UserTypeCtxt
  = ...
  | ThBrackCtxt         -- Template Haskell type brackets [t| ... |]
  | ...

However, this does not appear to actually be used anywhere and can be removed relatively easily:

diff --git a/compiler/GHC/Tc/Gen/HsType.hs b/compiler/GHC/Tc/Gen/HsType.hs
index 3a1ab09c9b..e81a87c3a8 100644
--- a/compiler/GHC/Tc/Gen/HsType.hs
+++ b/compiler/GHC/Tc/Gen/HsType.hs
@@ -2831,7 +2831,6 @@ expectedKindInCtxt :: UserTypeCtxt -> ContextKind
 -- Depending on the context, we might accept any kind (for instance, in a TH
 -- splice), or only certain kinds (like in type signatures).
 expectedKindInCtxt (TySynCtxt _)   = AnyKind
-expectedKindInCtxt ThBrackCtxt     = AnyKind
 expectedKindInCtxt (GhciCtxt {})   = AnyKind
 -- The types in a 'default' decl can have varying kinds
 -- See Note [Extended defaults]" in GHC.Tc.Utils.Env
diff --git a/compiler/GHC/Tc/Types/Origin.hs b/compiler/GHC/Tc/Types/Origin.hs
index b2fdefa4cd..cc63f7a5c5 100644
--- a/compiler/GHC/Tc/Types/Origin.hs
+++ b/compiler/GHC/Tc/Types/Origin.hs
@@ -92,7 +92,6 @@ data UserTypeCtxt
                         --    True:  stand-alone deriving
                         --    False: vanilla instance declaration
   | SpecInstCtxt        -- SPECIALISE instance pragma
-  | ThBrackCtxt         -- Template Haskell type brackets [t| ... |]
   | GenSigCtxt          -- Higher-rank or impredicative situations
                         -- e.g. (f e) where f has a higher-rank type
                         -- We might want to elaborate this
@@ -136,7 +135,6 @@ pprUserTypeCtxt (StandaloneKindSigCtxt n) = text "a standalone kind signature fo
 pprUserTypeCtxt TypeAppCtxt       = text "a type argument"
 pprUserTypeCtxt (ConArgCtxt c)    = text "the type of the constructor" <+> quotes (ppr c)
 pprUserTypeCtxt (TySynCtxt c)     = text "the RHS of the type synonym" <+> quotes (ppr c)
-pprUserTypeCtxt ThBrackCtxt       = text "a Template Haskell quotation [t|...|]"
 pprUserTypeCtxt PatSigCtxt        = text "a pattern type signature"
 pprUserTypeCtxt ResSigCtxt        = text "a result type signature"
 pprUserTypeCtxt (ForSigCtxt n)    = text "the foreign declaration for" <+> quotes (ppr n)
diff --git a/compiler/GHC/Tc/Validity.hs b/compiler/GHC/Tc/Validity.hs
index fba45562b7..05dcb28643 100644
--- a/compiler/GHC/Tc/Validity.hs
+++ b/compiler/GHC/Tc/Validity.hs
@@ -372,7 +372,6 @@ checkValidType ctxt ty

                  ForSigCtxt _   -> rank1
                  SpecInstCtxt   -> rank1
-                 ThBrackCtxt    -> rank1
                  GhciCtxt {}    -> ArbitraryRank

                  TyVarBndrKindCtxt _ -> rank0
@@ -500,7 +499,6 @@ vdqAllowed :: UserTypeCtxt -> Bool
 vdqAllowed (KindSigCtxt {}) = True
 vdqAllowed (StandaloneKindSigCtxt {}) = True
 vdqAllowed (TySynCtxt {}) = True
-vdqAllowed (ThBrackCtxt {}) = True
 vdqAllowed (GhciCtxt {}) = True
 vdqAllowed (TyVarBndrKindCtxt {}) = True
 vdqAllowed (DataKindCtxt {}) = True
@@ -1333,7 +1331,6 @@ okIPCtxt ResSigCtxt             = True
 okIPCtxt GenSigCtxt             = True
 okIPCtxt (ConArgCtxt {})        = True
 okIPCtxt (ForSigCtxt {})        = True  -- ??
-okIPCtxt ThBrackCtxt            = True
 okIPCtxt (GhciCtxt {})          = True
 okIPCtxt SigmaCtxt              = True
 okIPCtxt (DataTyCtxt {})        = True

Is there a good reason to keep this around?