I'm using StableNames to have a notion of function equality, and I'm running into problems when using monadic functions.
Consider the code below, file Test.hs import System.Mem.StableName
import Control.Monad.State eq :: a -> b -> IO Bool eq a b = do pa <- makeStableName a pb <- makeStableName b return (hashStableName pa == hashStableName pb)
successor :: (Num a, Monad m) => a -> m a successor n = return (n+1) main :: IO () main = do b1 <- eq (successor :: Int -> Maybe Int) (successor :: Int -> Maybe Int)
b2 <- eq (successor :: Int -> State Int Int) (successor :: Int -> State Int Int) print (show b1 ++ " " ++ show b2) Running the code into ghci the result is "False False". There is some old post saying that this is due to the dictionary-passing style for typeclasses, and compiling with optimizations improves the situation.
Compiling with ghc --make -O Tests.hs and running the program, the result is "True True", which is what I expect. However, if I change main to be like the following: main :: IO () main = do b2 <- eq (successor :: Int -> State Int Int) (successor :: Int -> State Int Int) b1 <- eq (successor :: Int -> Maybe Int) (successor :: Int -> Maybe Int)
print (show b1 ++ " " ++ show b2) i.e. just changing the sequential order, and then compiling again with the same command, I get "True False", which is very confusing for me.
Similar situations happens when using the state monad transformer, and manually built variations of it. It sounds the problem is with hidden closures created somewhere that do not point to the same memory locations, so StableNames yields false for that cases, but it is not clear to me under what circumstances this situation happens. Is there other way to get some approximation of function equality? or a way to "configure" the behavior of StableNames in presence of class constraints?
I'm using the latests Haskell Platform on OS X Lion, btw. Thanks Ismael _______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
Thanks Lorenzo, I'm cc'ing the list with your response also:
As you point out, when you do some kind of "let-binding", using the where clause, or explicit let as in: main :: IO () main = do let f1 = (successor :: Int -> State Int Int) let f2 = (successor :: Int -> Maybe Int) b2 <- eq f2 f2 b1 <- eq f1 f1
print (show b1 ++ " " ++ show b2) The behavior is as expected. I guess the binding triggers some internal optimization or gives more information to the type checker; but I'm still not clear why it is required to be done this way -- having to let-bind every function is kind of awkward.
I know the details of StableNames are probably implementation-dependent, but I'm still wondering about how to detect / restrict this situation. Thanks
2012/6/26 Lorenzo Bolla <[hidden email]> From StableName docs: Ismael _______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
The point I was making is that StableName might be what you want. You are using it to check if two functions are the same by comparing their "stablehash". But from StableName documentation:
The reverse is not necessarily true: if two stable names are not equal, then the objects they name may still be equal. The `eq` you implemented means this, I reckon: if `eq` returns True then the 2 functions are equal, if `eq` returns False then you can't tell!
Does it make sense? L. On Tue, Jun 26, 2012 at 1:54 PM, Ismael Figueroa Palet <[hidden email]> wrote: Thanks Lorenzo, I'm cc'ing the list with your response also: _______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
2012/6/26 Lorenzo Bolla <[hidden email]> The point I was making is that StableName might be what you want. You are using it to check if two functions are the same by comparing their "stablehash". But from StableName documentation: Yes it does make sense, and I'm wondering why the hash are equal in one case but are not equal on the other case (i.e. using let/where vs not using it) because I'd like it to behave the same in both situations
Thanks again
Ismael _______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
I think about StableName like the "&" operator in C, that returns you the memory address of a variable. It's not the same for many reasons, but by analogy, if &x == &y then x == y, but &x != &y does not imply x != y.
So, values that are semantically equal, may be stored in different memory locations and have different StableNames. The fact that changing the order of the lines also changes the result of the computation is obviously stated in the type signature of makeStableName, which lives in the IO monad. On the other hand hashStableNAme is a pure function.
L. On Tue, Jun 26, 2012 at 3:26 PM, Ismael Figueroa Palet <[hidden email]> wrote:
_______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
In other words there is a difference between Identity and Equivalence. What you have implemented with StableName is an "Identity" (sometimes called "reference equality"), as opposed to an "Equivalence" (aka "value equality").
In Python, for example: >>> x = {1:2} >>> y = {1:2} >>> x == y True >>> x is y False
L. On Tue, Jun 26, 2012 at 3:42 PM, Lorenzo Bolla <[hidden email]> wrote: I think about StableName like the "&" operator in C, that returns you the memory address of a variable. It's not the same for many reasons, but by analogy, if &x == &y then x == y, but &x != &y does not imply x != y. _______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
thanks again for your comments, any idea on how to implement "Equivalence" for functions? 2012/6/26 Lorenzo Bolla <[hidden email]> In other words there is a difference between Identity and Equivalence. What you have implemented with StableName is an "Identity" (sometimes called "reference equality"), as opposed to an "Equivalence" (aka "value equality"). Ismael _______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
This is very tricky and it really depends on what you mean...
Formally, two functions are the same if they have the same domain and f(x) == g(x) for each x in the domain. But this is not always easy/feasible/efficient to implement! (See also http://en.wikipedia.org/wiki/Rice%27s_theorem and http://stackoverflow.com/questions/4844043/are-two-functions-equal.)
Depending on your problem, you might get away with just defining a "signature" of your function and compare them: for example the signature could be the concat of the function name, args types, etc. But I'm speculating here...
L. On Tue, Jun 26, 2012 at 4:50 PM, Ismael Figueroa Palet <[hidden email]> wrote:
_______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
Yes I agree, so far StableNames have been a good approximation, except for the problem I described when using monads/class constraints :-(
2012/6/26 Lorenzo Bolla <[hidden email]> This is very tricky and it really depends on what you mean... Ismael _______________________________________________ Haskell-Cafe mailing list [hidden email] http://www.haskell.org/mailman/listinfo/haskell-cafe |
Powered by Nabble | Edit this page |