Hi Colin The following code probably would do what I wanted to do. play_move :: IORef Game_state -> IO () play_move game_state_ior = do (_, state, _) <- readIORef game_state_ior putStr "Playing AI: " start_time <- getCurrentTime let move = recommended_move state end_time <- move `seq` (modifyIORef game_state_ior $! update_interactive_from_move move) `seq` getCurrentTime putStrLn $ show $ (diffUTCTime end_time start_time) Due to the non-strictness of Haskell, the evaluation of the expressions between start_time and end_time is deferred until there is a need. By using `seq` and ($!), the strictness can be forced. Sean On Wed, Mar 18, 2009 at 9:28 PM, Colin Paul Adams wrote:

...

...

...

...

...

"Yitzchak" == Yitzchak Gale writes:

   Yitzchak> Colin Paul Adams wrote:    >> The code of the following routine is intended to indicate how    >> long it takes for the computer to make a move. However the time    >> is printed (as very close to zero) long before the move is    >> made.

   Yitzchak> You are writing a thunk to the IORef. It only gets    Yitzchak> computed later on when you read the value.

   Yitzchak> Try using:

   >> readIORef game_state_ior >>= evaluate    >> . update_interactive_from_move move >>= writeIORef    >> game_state_ior

   Yitzchak> evaluate is in Control.Exception.

Still no joy with:

play_move :: IORef Game_state -> IO () play_move game_state_ior = do  (_, state, _) <- readIORef game_state_ior  putStr "Playing AI: "  start_time <- getCurrentTime  let move = recommended_move state  readIORef game_state_ior >>=    evaluate . update_interactive_from_move move >>=             writeIORef game_state_ior  end_time <- getCurrentTime  putStrLn $ show $ (diffUTCTime end_time start_time)

-- Colin Adams Preston Lancashire _______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

-- Sean Lee PhD Student Programming Language and Systems Research Group School of Computer Science and Engineering University of New South Wales http://www.cse.unsw.edu.au/~seanl

You get some random values? or no change at all on the state? On Wed, Mar 18, 2009 at 9:45 PM, Colin Paul Adams wrote:

...

...

...

...

...

"Sean" == Sean Lee writes:

   Sean> Hi Colin The following code probably would do what I wanted    Sean> to do.

   Sean> play_move :: IORef Game_state -> IO () play_move    Sean> game_state_ior = do (_, state, _) <- readIORef    Sean> game_state_ior putStr "Playing AI: " start_time <-    Sean> getCurrentTime let move = recommended_move state end_time <-    Sean> move `seq` (modifyIORef game_state_ior $!    Sean> update_interactive_from_move move) `seq` getCurrentTime    Sean> putStrLn $ show $ (diffUTCTime end_time start_time)

   Sean> Due to the non-strictness of Haskell, the evaluation of the    Sean> expressions between start_time and end_time is deferred    Sean> until there is a need.  By using `seq` and ($!), the    Sean> strictness can be forced.

It certainly causes a time delay, and a non-zero time is printed, but the state doesn't get modified correctly now.

??

I will ponder Daniel's suggestion. -- Colin Adams Preston Lancashire

-- Sean Lee PhD Student Programming Language and Systems Research Group School of Computer Science and Engineering University of New South Wales http://www.cse.unsw.edu.au/~seanl

Am Mittwoch, 18. März 2009 11:54 schrieb Sean Lee:

You get some random values? or no change at all on the state?

On Wed, Mar 18, 2009 at 9:45 PM, Colin Paul Adams

wrote:

...

...

...

...

...

> "Sean" == Sean Lee writes:

Sean> Hi Colin The following code probably would do what I wanted Sean> to do.

Sean> play_move :: IORef Game_state -> IO () play_move Sean> game_state_ior = do (_, state, _) <- readIORef Sean> game_state_ior putStr "Playing AI: " start_time <- Sean> getCurrentTime let move = recommended_move state end_time <- Sean> move `seq` (modifyIORef game_state_ior $! Sean> update_interactive_from_move move) `seq` getCurrentTime

that should probably have been move `seq` (modifyIORef game_state_ior $! ... ) >> getCurrentTime value `seq` ioAction1 `seq` ioAction2 doesn't execute ioAction1.

...

Sean> putStrLn $ show $ (diffUTCTime end_time start_time)

Sean> Due to the non-strictness of Haskell, the evaluation of the Sean> expressions between start_time and end_time is deferred Sean> until there is a need. By using `seq` and ($!), the Sean> strictness can be forced.

It certainly causes a time delay, and a non-zero time is printed, but the state doesn't get modified correctly now.

??

I will ponder Daniel's suggestion. -- Colin Adams Preston Lancashire

The following code probably would do what I wanted to do.

I mean "what you wanted to do." :) -- Sean Lee PhD Student Programming Language and Systems Research Group School of Computer Science and Engineering University of New South Wales http://www.cse.unsw.edu.au/~seanl

Am Mittwoch, 18. März 2009 11:28 schrieb Colin Paul Adams:

...

...

...

...

...

"Yitzchak" == Yitzchak Gale writes:

Yitzchak> Colin Paul Adams wrote: >> The code of the following routine is intended to indicate how >> long it takes for the computer to make a move. However the time >> is printed (as very close to zero) long before the move is >> made.

Yitzchak> You are writing a thunk to the IORef. It only gets Yitzchak> computed later on when you read the value.

Yitzchak> Try using: >> readIORef game_state_ior >>= evaluate >> . update_interactive_from_move move >>= writeIORef >> game_state_ior

Yitzchak> evaluate is in Control.Exception.

Still no joy with:

play_move :: IORef Game_state -> IO () play_move game_state_ior = do (_, state, _) <- readIORef game_state_ior putStr "Playing AI: " start_time <- getCurrentTime let move = recommended_move state readIORef game_state_ior >>= evaluate . update_interactive_from_move move >>= writeIORef game_state_ior end_time <- getCurrentTime putStrLn $ show $ (diffUTCTime end_time start_time)

You would have to completely evaluate move. Depending on what type move is, that might require much more than what seq or evaluate do. Perhaps import Control.Parallel.Strategies let move = ... (rnf move) `seq` mofifyIORef (...) ?