Sergey Perminov wrote:

I wished to get output of unix commands in haskell code.

So i wrote: ---------------------------------------------------------------------------------- import System.IO import System.Process

eval :: String -> IO String eval s = do (_,hOutput,_,hProcess) <- runInteractiveCommand s sOutput <- hGetContents hOutput waitForProcess hProcess return sOutput ----------------------------------------------------------------------------------

'eval' works well if output of evaluated command is less than 64Kb. If not - 'eval' never ends.

What may cause this problem?

The laziness of hGetContents does. As long as nothing needs sOutput, it will not be read, the process becomes stuck on a clogged pipe, waitForProcess does not return and therefore nobody needs sOutput. There are two ways out: you can leave the waitForProcess out and trust that it eventually terminates or you can make eval more strict:

eval :: String -> IO String eval s = do (_,hOutput,_,hProcess) <- runInteractiveCommand s sOutput <- hGetContents hOutput foldr seq (waitForProcess hProcess) sOutput return sOutput

...but you most certainly don't want to buffer large amounts of output in a String. Depending on your actual problem, there may be far better solutions, such as

eval :: String -> (String -> IO b) -> IO b eval s k = bracket (runInteractiveCommand s) (\(_,_,_,hProcess) -> waitForProcess hProcess) (\(_,output,_,_) -> hGetContents output >>= k)

You will still run into problems if the called process writes to stdout, though. -Udo -- "You live and learn. At any rate, you live." -- Marvin, the paranoid droid