- Haskell-Cafe - Haskell.org

Old Messages
by Mark Espinoza 17 Oct '03

17 Oct '03

>> A couple of days ago I posted a question about the cgi library, >> which was not answered. >Hopefully, this repeat request will get you some answers (though I note >that >Graham Klyne did answer your question at length on 23/6/2003). Is there a place where I can search through the old messages to see if someone has already answered my question? Thanks Mark

2 1

Re: "listProduct" -- is this a standard function?
by K. Fritz Ruehr 17 Oct '03

17 Oct '03

I don't know of a type-based search tool for Haskell libraries, but I vaguely recall someone pursuing this idea using type isomorphisms to provide more robust results. This actually came up at lunch the other day*, where someone pointed out that you'd want to allow not only for type isomorphisms (e.g., argument order), but also for functions which might fit your spec. when instantiated with (say) some default argument. I don't know if this latter idea was addressed in any of the "TI" work, eg Di Cosmo's book (see also Rittri, Thatte, and others). Note that much of the TI work mentions library search as a motivation--I thought that someone might actually have implemented something for Haskell, though. In any case, and until someone might fill in my vaque recall, my students have found the reference tool at zvon.org quite helpful: there is a type-based index of the standard libraries at: <http://zvon.org/other/haskell/Outputglobal/index.html> (but note that the class qualifiers are missing, so that sequence looks more general than it really is). They also have nice indices tied to the library structure, e.g., the Prelude index at: <http://zvon.org/other/haskell/Outputprelude/index.html> -- Fritz * Type isomorphisms don't often come up over lunch in my regular venue, but I am fortunate enough to be spending some sabbatical time with the PacSoft group at the Oregon Graduate Institute (highly recommended). PS: A quick googling suggests that perhaps I was remembering a paper by Rittri in JFP vol. 1, #1, which I don't have handy. Does anyone know if a practical tool was ever released? ----------------- Graham Klyne wrote: > At 22:04 15/10/03 -0500, Matt Harden wrote: >>On Wednesday 15 October 2003 11:07 am, Graham Klyne wrote: >> > I've constructed a "listProduct" function that I think I've seen >> somewhere >> > else... is it a standard function? If so, where is it defined? >> >>Yes. It's called "sequence". It's defined in the prelude. >>It works with arbitrary monads, not just lists. > > That's the one, thanks (both of you). > > I *thought* I'd seen this function before. > > I observe this seems to be a recurring problem for me... there are so many > very generic and useful data handling functions in Haskell, it's difficult > for a jobbing programmer (as opposed to a language expert) to be on top of > them all. And the very generic nature of these function means that it's > not so easy to index them (e.g. in the way that the Java Developers > Almanac > series does for Java). > > Am I overlooking any important developer resources? > > #g > > > ------------ > Graham Klyne > For email: > http://www.ninebynine.org/#Contact

1 0

parsing a file line by line
by Anagha 16 Oct '03

16 Oct '03

Hi , I am a new bie to haskell. I need to read a file and get it contents line by line and then eventually word by word. I have written the code to read a file , but the returned value is all the lines in the file. How do I parse this whole text message ( contents of the file) into individual lines ( that is parsing / splitting the whole text to get me one line at a time. Also once I get the individual line, I o parse it word by word. I have a start on this thing. by using the line function in haskell, but I have not managed to write it successfully. Can anyone help me in this. Below is the code to read a file. fileFun :: IO () fileFun = do putStr "Please input a file name: " fname <- getLine contents <- readFile fname How do I split the data returned in contents( Contents contains all the text in the file, which is given as input) Thanks Anagha

2 1

Cgi Library
by Mark Espinoza 15 Oct '03

15 Oct '03

Hello, I have ghc 5 and 6 installed on my redhat 7.3 system. I have downloaded and unpacked the cgi library, which wants to have hugs installed. Is there an easy way to make this work with ghc (including the examples, I think I'll need them), or should I just try to install hugs. Thank you. Sincerely, Mark

3 2

Efficient Matrix Library
by Shawn P. Garbett 15 Oct '03

15 Oct '03

I'm looking for an efficient matrix library for Haskell. Suggestions? Shawn

1 0

Proper Etiquette
by Mark Espinoza 15 Oct '03

15 Oct '03

Hello: A couple of days ago I posted a question about the cgi library, which was not answered. What are appropriate courses of action for me in a case like this? Thanks. Mark

3 2

fold{l|r} and short-circuit evaluation
by Graham Klyne 15 Oct '03

15 Oct '03

I'm trying to use foldl or foldr to run a computation over a list, in such a way that only as much of the list as may be needed is actually examined. I believe the 'and' function behaves like this (e.g. see testand below), but when I use a function that performs pairwise operations on the list elementsa, it seems that I always end up constructing the entire list, even if I don't need to examine the value of each element. I guess I could try and figure out my own 2-place folding function, but this seems as if there should be a common solution. My test code is below. The tests are evaluated as: test allSame1 or test allSame2 to use a foldl- or foldr-versions of the function respectively If I comment out the indefinitely long list case (test7) in the definition of 'test', the foldl version will pass all these test cases and foldr throws an error on test6. But if I include test7, using Hugs the foldr test case returns a C stack overflow, and the foldl case silently disappears. Is there a standard method for dealing with cases like this? (I ask out of curiosity rather than need: for my real application, I am using relatively short list with quite expensive combining operations, so I think the 'foldl' variant which doesn't actually examine list elements once the result is known, will work quite well for me.) #g -- -- Spike-allsame.hs -- -- Test use of folding functions to evaluate whether all members -- of a list are the same. -- -- What I'm really interested in is whether the list gets evaluated -- beyond what is needed to obtain the desired result. -- import Maybe ( Maybe(..), isJust ) test allSame = and [ test1 allSame , test2 allSame , test3 allSame , test4 allSame , test5 allSame , test6 allSame , test7 allSame ] test1 allSame = allSame [1,1,1,1] == True test2 allSame = allSame [1,2,3,4] == False test3 allSame = allSame [1,2,undefined,4] == False test4 allSame = allSame [1] == True test5 allSame = allSame [] == True test6 allSame = allSame [1,2,undefined] == False test7 allSame = allSame (1:2:repeat 3) == False testand = and (True:True:False:repeat True) longList :: [Int] longList = (1:2:repeat 3) allSame1 :: (Eq a) => [a] -> Bool allSame1 [] = True allSame1 [_] = True allSame1 (a:as) = isJust $ foldr nextSame1 (Just a) as nextSame1 :: (Eq a) => a -> Maybe a -> Maybe a nextSame1 _ Nothing = Nothing nextSame1 a1 (Just a) | a1 == a = Just a | otherwise = Nothing {- foldr f z [] = z foldr f z (x:xs) = f x (foldr f z xs) foldr nextSame (Just 1) [2,undefined,4] ..> nextSame 2 (foldr nextSame (Just 1) [undefined,4]) ..> nextSame 2 (nextSame undefined (foldr nextSame (Just 1) [4])) ..> nextSame 2 (nextSame undefined (nextSame 4 (foldr nextSame (Just 1) []))) ..> nextSame 2 (nextSame undefined (nextSame 4 ((Just 1)))) -} allSame2 :: (Eq a) => [a] -> Bool allSame2 [] = True allSame2 [_] = True allSame2 (a:as) = isJust $ foldl nextSame2 (Just a) as nextSame2 :: (Eq a) => Maybe a -> a -> Maybe a nextSame2 Nothing _ = Nothing nextSame2 (Just a) a1 | a1 == a = Just a | otherwise = Nothing {- foldl f z [] = z foldl f z (x:xs) = foldl f (f z x) xs foldl nextSame (Just 1) [2,undefined,4] ..> foldl nextSame (nextSame (Just 1) 2) [undefined,4] ..> foldl nextSame (nextSame (nextSame (Just 1) 2) undefined) [4] ..> foldl nextSame (nextSame (nextSame (nextSame (Just 1) 2) undefined) 4) [] ..> (nextSame (nextSame (nextSame (Just 1) 2) undefined) 4) -} ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact

4 5

Using field selectors in a type constructor
by Graham Klyne 15 Oct '03

15 Oct '03

I've run across a minor coding niggle on a couple opf accosions when using a type constructor with field selectors. The full code of my test case is below. The value 'test2' evaluates to True. The function that niggles me is this: [[ joinVarBindings :: (Eq a) => VarBinding a b -> VarBinding a b -> VarBinding a b joinVarBindings vb1 vb2 | vbNull vb1 = vb2 | vbNull vb2 = vb1 | otherwise = VarBinding { vbMap = mv12 , vbEnum = map (\v -> (v,fromJust (mv12 v))) $ boundVars vb1 `union` boundVars vb2 , vbNull = False } where mv12 = headOrNothing . filter isJust . flist [ vbMap vb1, vbMap vb2 ] ]] Is it really necessary to define mv12 as a separate "where" clause here? What I'd really like to do is assign it to field vbMap, and reference that from the definition of vbEnum, but I can't figure out if there's a way to do so. Writing this: [[ joinVarBindings vb1 vb2 | vbNull vb1 = vb2 | vbNull vb2 = vb1 | otherwise = VarBinding { vbMap = head . filter isJust . flist [ vbMap vb1, vbMap vb2 ] , vbEnum = map (\v -> (v,fromJust (vbMap v))) $ -- error here ^^^^^^^ boundVars vb1 `union` boundVars vb2 , vbNull = False } ]] Results in a fairly obvious type error: I'd need to have a way to say that vbMap is applied to the value under construction. Experience with Java would suggest maybe something like this: [[ , vbEnum = map (\v -> (v,fromJust (vbMap this v))) $ ]] but of course Haskell isn't Java. Does Haskell provide a way to do this, other than the use of a 'where' clause to factor out the common subexpression, as in my working code? #g -- -- spike-constructorfields.hs import Maybe ( isJust, fromJust ) import List ( union, find ) -- |VarBinding is the type of an arbitrary variable bindings -- value, where the type of the bound values is not specified. data VarBinding a b = VarBinding { vbMap :: a -> Maybe b , vbEnum :: [(a,b)] , vbNull :: Bool } -- |nullVarBinding: maps no query variables. nullVarBinding :: VarBinding a b nullVarBinding = VarBinding { vbMap = const Nothing , vbEnum = [] , vbNull = True } -- |Return a list of the variables bound by a supplied variable binding boundVars :: VarBinding a b -> [a] boundVars = map fst . vbEnum -- |Join a pair of query bindings, returning a new binding that -- maps all variables recognized by either of the input bindings. -- If the bindings should overlap, such overlap is not detected and -- the value from the first binding provided is used arbitrarily. joinVarBindings :: (Eq a) => VarBinding a b -> VarBinding a b -> VarBinding a b joinVarBindings vb1 vb2 | vbNull vb1 = vb2 | vbNull vb2 = vb1 | otherwise = VarBinding { vbMap = mv12 , vbEnum = map (\v -> (v,fromJust (mv12 v))) $ boundVars vb1 `union` boundVars vb2 , vbNull = False } where mv12 = headOrNothing . filter isJust . flist [ vbMap vb1, vbMap vb2 ] flist fs a = map ($ a) fs -- see also monad function 'ap' headOrNothing :: [Maybe a] -> Maybe a headOrNothing [] = Nothing headOrNothing (a:_) = a {- NOT this: -- |Join a pair of query bindings, returning a new binding that -- maps all variables recognized by either of the input bindings. -- If the bindings should overlap, such overlap is not detected and -- the value from the first binding provided is used arbitrarily. joinVarBindings :: (Eq a) => VarBinding a b -> VarBinding a b -> VarBinding a b joinVarBindings vb1 vb2 | vbNull vb1 = vb2 | vbNull vb2 = vb1 | otherwise = VarBinding { vbMap = head . filter isJust . flist [ vbMap vb1, vbMap vb2 ] , vbEnum = map (\v -> (v,fromJust (vbMap v))) $ boundVars vb1 `union` boundVars vb2 , vbNull = False } -} -- |Function to make a variable binding from a list of -- pairs of variable and corresponding assigned value. makeVarBinding :: (Eq a, Show a, Eq b, Show b) => [(a,b)] -> VarBinding a b makeVarBinding vrbs = if null vrbs then nullVarBinding else VarBinding { vbMap = selectFrom vrbs , vbEnum = vrbs , vbNull = null vrbs } where -- selectFrom bs is the VarBinding lookup function selectFrom :: (Eq a) => [(a,b)] -> a -> Maybe b selectFrom [] _ = Nothing selectFrom ((v,r):bs) l = if l == v then Just r else selectFrom bs l vb1 :: VarBinding Int String vb1 = makeVarBinding [(1,"a"),(2,"b"),(3,"c")] vb2 :: VarBinding Int String vb2 = makeVarBinding [(3,"cc"),(4,"dd"),(5,"ee")] vb12 = joinVarBindings vb1 vb2 test1 = vbEnum $ vb12 test2 = test1 == [(1,"a"),(2,"b"),(3,"c"),(4,"dd"),(5,"ee")] test3 = [test31,test32,test33,test34] test31 = vbMap vb12 1 -- Just "a" test32 = vbMap vb12 3 -- Just "c" test33 = vbMap vb12 5 -- Just "ee" test34 = vbMap vb12 0 -- Nothing ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact

4 4

putStr
by Jose Morais 14 Oct '03

14 Oct '03

Hi, I am trying to something like f1 :: Int -> Int f1 x = f2 x f2 :: Int -> Int f2 x = 2 * x but before f2 returns its result I'd like it to print something like "The initial value is " ++ show x. How could I do this? Thank you

5 6

RE: Using existential types
by oleg＠pobox.com 14 Oct '03

14 Oct '03

Tim Docker wrote: > I wish to define a flexible report, consisting of a set of columns. > The contents of each column can be defined by mapping a function to a > row descriptor, generating a value, and then converting each value > to a string. I also wish to present an aggregate value at the bottom > of the column defined by another function. > It seems that existential types are required to get this sort of > thing to work. Is there another way? It seems there is. Here's your code: > data ColDesc rowv = forall a. ColDesc (rowv -> a) ([a] -> a) (a -> > String) > calculate :: [rowv] -> ColDesc rowv -> ([String],String) > calculate rs (ColDesc valf sumf fmtf) = let vals = map valf rs in (map > fmtf vals, (fmtf.sumf) vals) Here's the new code: > data ColDesc rowv = ColDesc (rowv -> String) ([rowv] -> String) > calculate :: [rowv] -> ColDesc rowv -> ([String],String) > calculate rs (ColDesc fmtf sumf) = (map fmtf rs, sumf rs) > rows = ["I","wish","to","define","a","report"] > mapp g f = g . (map f) > cols = [ > ColDesc (id.id) $ id.(\_ -> ""), > ColDesc (show.length) $ show.(mapp sum length), > let valf = (\s -> length s * length s) in ColDesc (show.valf) $ show.(mapp maximum valf) > ] > values = [ calculate rows col | col <- cols ] It was derived with a simple transformation: composing (rowv -> a) with (a->String) to eliminate the existential type a. It seems there is a pattern here. When we write data Foo a b = forall ex. Foo <something1> <something2> <something3> with ex unconstrained, then one of <something> must be a function F(a,b) -> ex and another something is a function ex -> G(a,b). Here F(a,b) is some expression that includes ground types, a, b -- but not ex. Indeed, otherwise we have no way to convert something into ex and to get anything out. The third something is perhaps a transformer ex -> H(a,b) -> ex. It seems therefore we are guaranteed that there is always a composition of <something>s that does not include ex. Therefore, instead of carrying original <something>, we can carry the compositions, and get rid of the existential quantification. It seems that (g.f) :: i -> o is equivalent to stating exists c. f:: (i->c) ^ g::(c->o) or, in the Haskell type language data P i o = forall c. P (i->c) (c->o) P f g I also have a nagging feeling that this approach relates to Skolemization. Indeed, Mike Gunter's transformation from > data Stat i o = -- aggregate function taking i's on input and producing o > forall s. Stat > s -- init > (s -> i -> s) -- update > (s -> o) -- result to > data Stat i o = Stat { update :: i -> Stat i o > , result :: o } looks awfully like a Skolemization, with 'Stat i o' being the Skolem function.

4 8