Test Trac #10359

3 files changed, 135 insertions, 1 deletions

diff --git a/testsuite/tests/perf/should_run/T10359.hs b/testsuite/tests/perf/should_run/T10359.hs
new file mode 100644
index 0000000000..fa10560970
--- /dev/null
+++ b/testsuite/tests/perf/should_run/T10359.hs
@@ -0,0 +1,125 @@
+{-# LANGUAGE RankNTypes #-}
+{-# LANGUAGE RecordWildCards #-}
+{-# LANGUAGE ConstraintKinds #-}
+
+module Main( main, boo ) where
+
+import Prelude hiding (repeat)
+
+boo xs f = (\x -> f x, xs)
+
+repeat :: Int -> (a -> a) -> a -> a
+repeat 1 f x = f x
+repeat n f x = n `seq` x `seq` repeat (n-1) f $ f x
+
+---- Buggy version
+------------------
+
+type Numerical a = (Fractional a, Real a)
+
+data Box a = Box
+    { func :: forall dum. (Numerical dum) => dum -> a -> a
+    , obj :: !a }
+
+do_step :: (Numerical num) => num -> Box a -> Box a
+do_step number Box{..} = Box{ obj = func number obj, .. }
+
+start :: Box Double
+start = Box { func = \x y -> realToFrac x + y
+            , obj = 0 }
+
+test :: Int -> IO ()
+test steps = putStrLn $ show $ obj $ repeat steps (do_step 1) start
+
+---- Driver
+-----------
+
+main :: IO ()
+main = test 2000 -- compare test2 10000000 or test3 10000000, but test4 20000
+
+
+{-
+---- No tuple constraint synonym is better
+------------------------------------------
+
+data Box2 a = Box2
+    { func2 :: forall num. (Fractional num, Real num) => num -> a -> a
+    , obj2 :: !a }
+
+do_step2 :: (Fractional num, Real num) => num -> Box2 a -> Box2 a
+do_step2 number Box2{..} = Box2{ obj2 = func2 number obj2, ..}
+
+start2 :: Box2 Double
+start2 = Box2 { func2 = \x y -> realToFrac x + y
+              , obj2 = 0 }
+
+test2 :: Int -> IO ()
+test2 steps = putStrLn $ show $ obj2 $ repeat steps (do_step2 1) start2
+
+---- Not copying the function field works too
+---------------------------------------------
+
+do_step3 :: (Numerical num) => num -> Box a -> Box a
+do_step3 number b@Box{..} = b{ obj = func number obj }
+
+test3 :: Int -> IO ()
+test3 steps = putStrLn $ show $ obj $ repeat steps (do_step3 1) start
+
+---- But record wildcards are not at fault
+------------------------------------------
+
+do_step4 :: (Numerical num) => num -> Box a -> Box a
+do_step4 number Box{func = f, obj = x} = Box{ obj = f number x, func = f }
+
+test4 :: Int -> IO ()
+test4 steps = putStrLn $ show $ obj $ repeat steps (do_step4 1) start
+-}
+
+
+{-
+First of all, very nice example. Thank you for making it so small and easy to work with.
+
+I can see what's happening. The key part is what happens here:
+{{{
+do_step4 :: (Numerical num) => num -> Box a -> Box a
+do_step4 number Box{ func = f, obj = x}
+              = Box{ func = f, obj = f number x }
+}}}
+After elaboration (ie making dictionaries explicit) we get this:
+{{{
+do_step4 dn1 number (Box {func = f, obj = x })
+  = Box { func = \dn2 -> f ( case dn2 of (f,r) -> f
+                           , case dn2 of (f,r) -> r)
+        , obj = f dn1 number x }
+}}}
+That's odd!  We expected this:
+{{{
+do_step4 dn1 number (Box {func = f, obj = x })
+  = Box { func = f
+        , obj = f dn1 number x }
+}}}
+And indeed, the allocation of all those `\dn2` closures is what is causing the problem.
+So we are missing this optimisation:
+{{{
+   (case dn2 of (f,r) -> f, case dn2 of (f,r) -> r)
+===>
+   dn2
+}}}
+If we did this, then the lambda would look like `\dn2 -> f dn2` which could eta-reduce to `f`.
+But there are at least three problems:
+ * The tuple transformation above is hard to spot
+ * The tuple transformation is not quite semantically right; if `dn2` was bottom, the LHS and RHS are different
+ * The eta-reduction isn't quite semantically right: if `f` ws bottom, the LHS and RHS are different.
+
+You might argue that the latter two can be ignored because dictionary arguments are special;
+indeed we often toy with making them strict.
+
+But perhaps a better way to avoid the tuple-transformation issue would be not to construct that strange expression in the first place. Where is it coming from?  It comes from the call to `f` (admittedly applied to no arguments) in `Box { ..., func = f }`.  GHC needs a dictionary for `(Numerical dum)` (I changed the name of the type variable in `func`'s type in the definition of `Box`).  Since it's just a pair GHC says "fine, I'll build a pair, out of `Fractional dum` and `Real dum`.  How does it get those dictionaries?  By selecting the components of the `Franctional dum` passed to `f`.
+
+If GHC said instead "I need `Numerical dum` and behold I have one in hand, it'd be much better. It doesn't because tuple constraints are treated specially.  But if we adopted the idea in #10362, we would (automatically) get to re-use the `Numerical dum` constraint.  That would leave us with eta reduction, which is easier.
+
+As to what will get you rolling, a good solution is `test3`, which saves instantiating and re-generalising `f`. The key thing is to update all the fields ''except'' the polymorphic `func` field. I'm surprised you say that it doesn't work.  Can you give a (presumably more complicated) example to demonstrate? Maybe there's a separate bug!
+
+-}
+
+
diff --git a/testsuite/tests/perf/should_run/T10359.stdout b/testsuite/tests/perf/should_run/T10359.stdout
new file mode 100644
index 0000000000..f6f4e0735a
--- /dev/null
+++ b/testsuite/tests/perf/should_run/T10359.stdout
@@ -0,0 +1 @@
+2000.0
diff --git a/testsuite/tests/perf/should_run/all.T b/testsuite/tests/perf/should_run/all.T
index f6801040e3..c95dfa0110 100644
--- a/testsuite/tests/perf/should_run/all.T
+++ b/testsuite/tests/perf/should_run/all.T
@@ -1,8 +1,16 @@
 # Tests that newArray/newArray_ is being optimised correctly
 
+test('T10359',
+     [stats_num_field('bytes allocated',
+          [(wordsize(64), 499512, 5),
+           (wordsize(32), 250000, 5)]),
+      only_ways(['normal'])
+      ],
+     compile_and_run,
+     ['-O'])
+
 # fortunately the values here are mostly independent of the wordsize,
 # because the test allocates an unboxed array of doubles.
-
 test('T3586',
      [stats_num_field('peak_megabytes_allocated', (17, 1)),
                                  # expected value: 17 (amd64/Linux)