Ticket #4229 (closed bug: fixed)
relatively simple test case for "internal error: PAP object entered!" on GHC 6.12.3
|Reported by:||carlhowells||Owned by:|
|Type of failure:||None/Unknown||Difficulty:|
|Test Case:||Blocked By:||#3983|
Description (last modified by igloo) (diff)
So, here's a test case
-- bug.cabal Name: bug Version: 0.1 Build-type: Simple Cabal-version: >=1.2 Executable bug Main-is: Bug.hs Build-depends: base > 4 && < 5, HUnit >= 188.8.131.52 && < 1.3, monads-fd >= 0.1 && < 0.2
-- Bug.hs module Main where import Control.Monad.Trans import Control.Monad.State.Strict import Test.HUnit runMyTest :: StateT () IO a -> IO a runMyTest x = evalStateT x () explode :: StateT () IO () explode = when ("" == "") . liftIO . assertFailure $ "X" main :: IO Counts main = runTestTT . TestCase . runMyTest $ explode
$ cabal build && ./dist/build/bug/bug Preprocessing executables for bug-0.1... Building bug-0.1... [1 of 1] Compiling Main ( Bug.hs, dist/build/bug/bug-tmp/Main.o ) Linking dist/build/bug/bug ... Cases: 1 Tried: 0 Errors: 0 Failures: 0bug: internal error: PAP object entered! (GHC version 6.12.3 for x86_64_unknown_linux) Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug Aborted
I couldn't figure out how to get the error to happen with HUnit or a monad transformer. But here are some additional notes:
Compiling with -O0 fixes it.
Using mtl instead of monads-fd doesn't change anything.
Changing the ("" == "") to True results in not getting a GHC internal error, but the test run reports 0 failures, which is incorrect. Changing the import of Control.Monad.State.Strict to Control.Monad.State also results in no crash, but 0 failures reported.
None of these issues occurs in GHC 6.10.4. I haven't tested on versions other than 6.10.4 and 6.12.3.