#3542: ghc-cabal deadlocks
---------------------------------+------------------------------------------
Reporter: igloo | Owner:
Type: bug | Status: new
Priority: high | Milestone: 6.12.1
Component: Build System | Version: 6.10.4
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Os: Unknown/Multiple
Architecture: Unknown/Multiple |
---------------------------------+------------------------------------------
Comment (by simonmar):
I mentioned this to Ian this morning on the phone, but just to record it
here: it is indeed possible to write this code correctly using the single-
threaded RTS, an example of how to do so is the implementation of
`readProcess` and `readProcessWithExitCode` in recent versions of
`System.Process`. Here's a snippet:
{{{
readProcess cmd args input = do
(Just inh, Just outh, _, pid) <-
createProcess (proc cmd args){ std_in = CreatePipe,
std_out = CreatePipe,
std_err = Inherit }
-- fork off a thread to start consuming the output
output <- hGetContents outh
outMVar <- newEmptyMVar
_ <- forkIO $ C.evaluate (length output) >> putMVar outMVar ()
-- now write and flush any input
when (not (null input)) $ do hPutStr inh input; hFlush inh
hClose inh -- done with stdin
-- wait on the output
takeMVar outMVar
hClose outh
-- wait on the process
ex <- waitForProcess pid
...
}}}
the idea is that you block on an `MVar` instead of in `waitForProcess`.
This assumes that if the process goes away then the pipe will also go away
and the threads will stop. Actually I think the code should be using
`finally` to make sure the `MVar` gets written even if an exception is
raised.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/3542#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
