Pmc: consider any 2 dicts of the same type equal

This patch massages the keys used in the `TmOracle` `CoreMap` to ensure
that dictionaries of coherent classes give the same key.
That is, whenever we have an expression we want to insert or lookup in
the `TmOracle` `CoreMap`, we first replace any dictionary
`$dict_abcd :: ct` with a value of the form `error @ct`.

This allows us to common-up view pattern functions with required
constraints whose arguments differed only in the uniques of the
dictionaries they were provided, thus fixing #21662.

This is a rather ad-hoc change to the keys used in the
`TmOracle` `CoreMap`. In the long run, we would probably want to use
a different representation for the keys instead of simply using
`CoreExpr` as-is. This more ambitious plan is outlined in #19272.

Fixes #21662
Updates unix submodule

3 files changed, 22 insertions, 8 deletions

diff --git a/testsuite/tests/pmcheck/should_compile/T11822.stderr b/testsuite/tests/pmcheck/should_compile/T11822.stderr
index 7f0aae9aec..d2994b65aa 100644
--- a/testsuite/tests/pmcheck/should_compile/T11822.stderr
+++ b/testsuite/tests/pmcheck/should_compile/T11822.stderr
@@ -12,11 +12,3 @@ T11822.hs:33:1: warning: [-Wincomplete-patterns (in -Wextra)]
             _ _
             _ (Data.Sequence.Internal.Seq Data.Sequence.Internal.EmptyT) _ _
             ...
-
-T11822.hs:33:1: warning:
-    Pattern match checker ran into -fmax-pmcheck-models=30 limit, so
-      • Redundant clauses might not be reported at all
-      • Redundant clauses might be reported as inaccessible
-      • Patterns reported as unmatched might actually be matched
-    Suggested fix:
-      Increase the limit or resolve the warnings to suppress this message.
diff --git a/testsuite/tests/pmcheck/should_compile/T21662.hs b/testsuite/tests/pmcheck/should_compile/T21662.hs
new file mode 100644
index 0000000000..1a3ff6cb6e
--- /dev/null
+++ b/testsuite/tests/pmcheck/should_compile/T21662.hs
@@ -0,0 +1,21 @@
+{-# LANGUAGE AllowAmbiguousTypes #-}
+{-# LANGUAGE ScopedTypeVariables #-}
+{-# LANGUAGE ViewPatterns #-}
+{-# LANGUAGE DataKinds #-}
+{-# LANGUAGE PolyKinds #-}
+{-# LANGUAGE TypeApplications #-}
+
+module T21662 where
+
+import GHC.TypeNats (Nat, KnownNat)
+
+view_fn :: forall (n :: Nat). KnownNat n => Int -> Bool
+view_fn i = i > 0
+
+foo :: Int -> Int
+foo (view_fn @12 -> True) = 0
+foo (view_fn @12 -> False) = 0
+
+  -- The point is that the two view pattern functions "view_fn"
+  -- may get different uniques for the KnownNat 12 dictionary,
+  -- which leads to a spurious pattern match warning.
diff --git a/testsuite/tests/pmcheck/should_compile/all.T b/testsuite/tests/pmcheck/should_compile/all.T
index fe63d364f7..6415e83dab 100644
--- a/testsuite/tests/pmcheck/should_compile/all.T
+++ b/testsuite/tests/pmcheck/should_compile/all.T
@@ -111,6 +111,7 @@ test('CyclicSubst', [],  compile, [overlapping_incomplete])
 test('CaseOfKnownCon', [], compile, [overlapping_incomplete])
 test('TooManyDeltas', [], compile, [overlapping_incomplete+'-fmax-pmcheck-models=0'])
 test('LongDistanceInfo', [], compile, [overlapping_incomplete])
+test('T21662', [],  compile, [overlapping_incomplete])
 
 # Series (inspired) by Luke Maranget