Re: [Haskell-cafe] Set of reals...?

erm, yes you're right - don't know why that is - seems a fairly arbitrary decision to me... perhaps someone else knows a good reason why normal function definiton is not allowed? Stijn De Saeger wrote:

aha, I see. Seems like i still have a long way to go with functional programming.

final question: i tried to test the code below, but it seems GHCi will only take the `isin` functions when they are defined in lambda notation (like isin = (\x -> ...)). Did you run this code too, or were you just sketching me the rough idea?

Cheers for all the replies by the way, i learnt a great deal here. stijn.

On Wed, 27 Oct 2004 14:09:36 +0100, Keean Schupke wrote:

...

Well, its functional of course:

union :: Interval -> Interval -> Interval union i j = Interval { isin x = isin i x || isin j x }

intersection :: Interval -> Interval -> Interval intersection i j = Interval { isin x = isin i x && isin j x }

Keean.

Stijn De Saeger wrote:

...

That seems like a very clean way to define the sets indeed, but how would you go about implementing operations like intersection, complement etc... on those structures? define some sort of algebra over the functions? or extend such sets by adding elements? hm... sounds interesting,.

thanks, stijn.

On Wed, 27 Oct 2004 11:52:54 +0100, Keean Schupke wrote:

...

I think someone else mentioned using functions earlier, rather than a datatype why not define:

data Interval = Interval { isin :: Float -> Bool }

Then each range becomes a function definition, for example:

myInterval = Interval { isin r | r == 0.6 = True | r > 0.7 && r < 1.0 = True | otherwise = False }

Then you can test with:

(isin myInterval 0.6)

Keean