testsuite/process011: Don't rely on Python

[bgamari]

The refactor of process011 performed in 8cd7e04, where the test is taught to use python3 instead of sh, has broken the test on Debian 10. Specifically, for reasons that I don't yet understand, python3 exits with code 1 not code 2 when killing itself with signal 2. Strangely, my tests on other distributions (e.g. NixOS) suggest that Python's behavior in this case is rather inconsistent.

To avoid this, we now rather use a dedicated C program instead of Python.

Fixes #241.

[bgamari]

I am validating this in https://gitlab.haskell.org/ghc/ghc/-/merge_requests/7722.

[blackgnezdo]

Better yet, instead of exec, apply this patch:

diff --git a/tests/process011.hs b/tests/process011.hs
index 83e2030..b711fe4 100644
--- a/tests/process011.hs
+++ b/tests/process011.hs
@@ -15,7 +15,7 @@ main = do
   -- shell kills itself with SIGINT,
   -- delegation off, exit code (death by signal) reported as normal
   do let script = "./process011_c"
-     (_,_,_,p) <- createProcess (shell script) { delegate_ctlc = False }
+     (_,_,_,p) <- createProcess (proc script []) { delegate_ctlc = False }
      waitForProcess p >>= print

   putStrLn "===================== test 2"
@@ -23,7 +23,7 @@ main = do
   -- shell kills itself with SIGINT,
   -- delegation on, so expect to throw UserInterrupt
   do let script = "./process011_c"
-     (_,_,_,p) <- createProcess (shell script) { delegate_ctlc = True }
+     (_,_,_,p) <- createProcess (proc script []) { delegate_ctlc = True }
      (waitForProcess p >>= print)
        `catchUserInterrupt` \e -> putStrLn $ "caught: " ++ show e

[bgamari]

Thanks @blackgnezdo !