Depending on the kind of state, WHNF may not be enough - have you tried with deepseq?

On Sun, Jan 12, 2014 at 5:07 PM, Corentin Dupont <corentin.dupont@gmail.com> wrote:

Hi guys,

I'm experimenting different behaviours with or without a "putStrLn"! :(

Basically, with the following code, I want the evaluation to really happen on the "evaluate".

I found out that it doesn't: it is evaluated elsewhere (I don't know where).

If I put a putStrLn (commented below), the evaluation really happens there.

execCommand :: (TVar MyState) -> StateT MyState IO () -> IO ()
execCommand ts sm = do
   s <- atomically $ readTVar ts
   s' <- execStateT sm s
   s'' <- evaluate s'          --evaluation should happen here, but it doesn't
   --putStrLn $ displayMulti $ _multi s''
   atomically $ writeTVar ts s''

To give you more context, I have a state that, when evaluated, might not terminate.
So I added a watchdog (like in mueval), that will kill the thread in case the evaluation doesn't terminate.

That's why I need to be sure of where the evaluation takes place.

Thanks!

Corentin

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe