Hi Guys, thanks again for the time reading this thread.
I was busy so I couldn't modify the code, but now it looks like this:
main = do
finished <- atomically $ newTChan
chan <- atomically $ newTChan
f <- openOffline "filename"
forkIO $ writeFile chan
forkIO $ reaFile f chan finished
a <- readTChan finished
show a
So chan is the variable for communication and finished the variable used to readFile tell the main thread that the file is over,
It seems more neat than the previous code,
but I have a hard time understanding the sentence "...you have to manually specify the number of threads that you opened..." in Felipe's email. What does it mean? How do I specify automatically how many threads do I want?
Thanks a lot
Mau :)
On Mon, Jul 2, 2012 at 10:18 PM, Bob Hutchison
<hutch-lists@recursive.ca> wrote:
On 2012-07-02, at 9:10 AM, Bob Hutchison wrote:
On 2012-06-30, at 1:51 PM, Mauricio Hernandes wrote:
Eternal Gratitude for the help, it's working perfectly, I will consider the exceptions and other stuff now.
the code looks like this now
import System.IO
import Control.Concurrent
import Data.List
import Control.Monad
main = do
finished <- newEmptyMVar
input <- newMVar [1..30000]
ia <- newEmptyMVar
ib <- newEmptyMVar
ic <- newEmptyMVar
forkIO $ do x <- readMVar input
putMVar ia x
putMVar finished ()
forkIO $ do a <- readMVar ia
putMVar ib ( sum a )
putMVar finished ()
forkIO $ do a <- readMVar ia
putMVar ic ( reverse a )
putMVar finished ()
b <- readMVar ib
c <- readMVar ic
writeFile "somaEprod.txt" (show b ++ "\n")
appendFile "somaEprod.txt" (show c)
replicateM_ 3 (takeMVar finished)
Just another Haskell beginner here, so beware...
You've moved the readMVar out of a thread into the application. This means (I think) that you are waiting for values in both ib and ic in the application (rather than a fourth thread). In your specific program, for these to have values require that all three threads have completed so you don't need the finished MVar anymore. However, this is pretty fragile being completely dependent on the MVars being set exactly once as the threads complete (so if you modify the code you have to be careful). The found solution is also fragile as
^^^^^ finished (sorry)
Felipe says in his post. I don't know what those libraries Felipe mentioned are but I think I'd be looking for them right about now if I were you :-)
Cheers,
Bob
Valeu
Mauricio
On Sun, Jul 1, 2012 at 12:24 AM, Felipe Almeida Lessa
<felipe.lessa@gmail.com> wrote:
Your application is exiting before your forkIOs get a chance to
execute. Instead of
forkIO $ do
...
forkIO $ do
...
forkIO $ do
...
use something like
finished <- newEmptyMVar
forkIO $ do
...
putMVar finished ()
forkIO $ do
...
putMVar finished ()
forkIO $ do
...
putMVar finished ()
replicateM_ 3 (takeMVar finished)
Doing so will avoid your program to exit until all threads have finished.
Note that the code above is extremely fragile: doesn't handle
exceptions, you have to manually specify the number of threads that
you opened, etc. These are abstracted by some libraries on Hackage
that you may use later for Real World Code (TM).
Cheers, =)
--
Felipe.
_______________________________________________
Beginners mailing list
Beginners@haskell.org
http://www.haskell.org/mailman/listinfo/beginners
----
Bob Hutchison
Recursive Design Inc.
_______________________________________________
Beginners mailing list
Beginners@haskell.org
http://www.haskell.org/mailman/listinfo/beginners
----
Bob Hutchison
Recursive Design Inc.