[Haskell-cafe] Space questions about intern and sets

[Gracjan Polak]

Hello all,

I've got two questions, both are space related so I'll put them in one 
e-mail.

1. I'd like to have function intern, that behaves essentially like id, 
but with the following constraint:
    if x==y then makeStableName(intern x)==makeStableName(intern y)
Reasoning behind this hack: objects equal (as in Eq) should occupy the 
same place in memory. I've got to parse quite large file, most tokens 
are the same. Haskell speed is very good, but I constantly run out of 
memory. Here is something I wrote, but it doesn't work :(

intern :: Ord a => a -> a
intern x = unsafePerformIO $ internIO x

iorefset :: Ord a => IORef(Map.Map a a)
iorefset = unsafePerformIO $ do
     newIORef $ Map.empty

{-# NOINLINE iorefset #-}

internIO :: Ord a => a -> IO a
internIO x = do
     myset <- readIORef iorefset
     case Map.lookup x myset of
         Just y -> do
             return y
         Nothing -> do
             let newset = Map.insert x x myset
             writeIORef iorefset newset
             return $ fromJust $ Map.lookup x newset
             --return x

iorefset is re-executed, but it shouldn't.

2. Second question: imagine a Data.Set.Set and Data.Set.map with some 
function f, that is mostly identity, only some, rare elements are 
changed. As I understand it, map essentially copies whole structure of 
set and this leads to out-of-memory conditions for me.

So, basically I have space problems. Is there any FAQ out there "how to 
make my program fit in 256MB of heap?"

-- 
Gracjan