- Haskell - Haskell.org

Re: Ambiguous types
by Ashley Yakeley 21 Jan '02

21 Jan '02

At 2002-01-21 00:25, Janis Voigtlaender wrote: >Hope that helps and doesn't just make the story more obscure ;-) I'm not convinced. Here's a simplified case... This compiles fine: emptyList :: [a] emptyList = [] isempty :: Bool isempty = null emptyList But this gives an error: emptyList :: (Ord a) => [a] emptyList = [] isempty :: Bool isempty = null emptyList ghci says: Ambiguous type variable(s) `a' in the constraint `Ord a' arising from use of `emptyList' at ActorTest.hs:7 In the first argument of `null', namely `emptyList' In the definition of `isempty': null emptyList Hugs says: ERROR "ActorTest.hs" (line 7): Cannot justify constraints in explicitly typed binding *** Expression : isempty *** Type : Bool *** Given context : () *** Constraints : Ord a -- Ashley Yakeley, Seattle WA

2 1

Ambiguous types
by Daniel . 21 Jan '02

21 Jan '02

Hi there, I've just starting learning Haskell and, to be quite frank, it gives me a cold sweat at times. What has me confused at present is why the following code gives me :- ActorTest.hs:8: Ambiguous type variable(s) `a' in the constraint `Action a' arising from use of `validActions' at ActorTest.hs:8 In the first argument of `(==)', namely `validActions actor' In the definition of `canAct': (validActions actor) == [] {-code-} module ActorTest where class Actor t where canAct :: t -> Bool validActions :: (Action u) => t -> [u] canAct actor = validActions actor == [] validActions actor = [] class Eq t => Action t {-end code-} Is the error saying that it is impossible to tell what specific type instance of the Action class will be returned by validActions? If so, why should it matter? Isn't it enough to know that they results will be of the class Action? Thanks Daniel -- _______________________________________________ Sign-up for your own FREE Personalized E-mail at Mail.com http://www.mail.com/?sr=signup 1 cent a minute calls anywhere in the U.S.! http://www.getpennytalk.com/cgi-bin/adforward.cgi?p_key=RG9853KJ&url=http:/…

2 1

RE: dependent type query
by Simon Peyton-Jones 21 Jan '02

21 Jan '02

| >Furthermore, I can't even write: | > | >data (MyClass f a) => MyData2 f = MyData2 f | | Well now this is odd. I had no trouble simulating it with this: | | data MyData2 f = MyData2 f | | mkMyData2 :: (MyClass f a) => f -> MyData2 f | mkMyData2 = MyData2 | | Looks like extended Haskell is being excessively restricted. Comments? I think you are correct here. But as you point out, contexts in data declarations only restrict what programs you can write, by adding extra constraints to the constructors. They have no useful purpose at all, so far as I know, and I wish Haskell didn't have them. So I am disinclined to invest much effort in fixing this infelicity in GHC's handling of them. Simon

1 0

(no subject)
by Peter Stuckey 21 Jan '02

21 Jan '02

------- Blind-Carbon-Copy To: pjs Subject: Date: Mon, 21 Jan 2002 13:27:08 +1100 Message-ID: <21510.1011580028(a)muldi.cs.mu.oz.au> From: Peter Stuckey <pjs(a)muldi.cs.mu.oz.au> Apologies for Receiving Multiple Copies of this Message - ---------------------------------------------------------- CALL FOR PAPERS ICLP'02 Eighteenth International Conference on Logic Programming Copenhagen, Denmark July 29th - August 1, 2002 http://floc02.diku.dk/ICLP/ ICLP'02, the Eighteenth International Conference on Logic Programming will be be held this year as part of the 2002 Federated Logic Conference (http://floc02.diku.dk/) in conjunction with CADE, CAV, FME, LICS, RTA and TABLEAUX. ICLP'02 is sponsored by IF/Prolog (http://www.ifcomputer.de) TOPICS Since the first ICLP, held in Marseilles in 1982, ICLP has been the premier international conference for presenting research into logic programming. Original papers are sought in all areas of logic programming including (but not restricted to): Theory Implementation Semantic Foundations Compilation Formalisms Memory Management Non-monotonic Reasoning Virtual Machines Knowledge Representation Parallelism Language Issues Environments Constraints Program Analysis Concurrency Program Transformation Objects Validation and Verification Coordination Debugging Mobilility Higher Order Applications Types Modes Deductive Databases Programming Techniques Software Engineering Natural Language Web tools Internet Agents Artificial Intelligence Papers describing innovative applications of logic programming (in the broadest sense, e.g. including constraint programming, non-monotonic systems, etc.) are particularly sought. We welcome submission in, but not limited to, the following topics: surveys of an application area, problems to which logic programming may be applied, experience in applying logic programming, software engineering aspects of logic programming and areas where further research is required to meet industrial needs. There will be an award for the best application paper sponsored by IF/Prolog. SUBMISSION Papers must describe original, previously unpublished work, be written and presented in English, not exceed 15 pages (A4 or letter format, up to 5,000 words), and not be simultaneously submitted for publication elsewhere. The proceedings will be published by Springer Verlag in the Lecture Notes in Computer Science series. Authors are strongly encouraged to use LaTeX2e and the Springer llncs class file, available at http://www.springer.de/comp/lncs/authors.html Submission is Web-based. In order to submit a paper, authors should upload it via the web at http://www.cs.mu.oz.au/~pjs/ICLP2002/ where more detailed instructions are given. If submission through the Web is not possible, five hard copies may be sent to the program chair. Authors intending to submit a paper are requested to submit an abstract of their paper, through the same web interface, before February 10, to help expedite the paper assignment process. Submission Agenda: Submission of abstracts: February 10, 2002 Submission of papers: February 17, 2002 Notification of acceptance: April 14, 2002 Camera-ready papers due: May 14, 2002 Workshops: The following workshops will be held in conjunction with ICLP'02. August 1: CLIMA - Computational Logic In Multi-Agent Systems July 28: CLPSE - (Constraint) Logic Programming and Software Engineering July 31: LPE - Workshop on Logic Programming Environments July 27-28: NLULP - Natural Language Understanding and Logic Programming July 27: PCL - Paraconsistent Computational Logic July 27: SAVE - Specification, Analysis and Validation for Emerging Technologies in Computational Logic July 28: UNKB - Updating Non-Monotonic Knowledge Bases Conference Location: The conference will be held at the University of Copenhagen. Program Chair: Conference Chair: Peter J. Stuckey Henning Christiansen Department of Computer Science Department of Computer Science and Software Engineering Roskilde University University of Melbourne PO Box 260 3010, AUSTRALIA DK-4000 Roskilde, DENMARK Email: pjs(a)cs.mu.oz.au Email: henning(a)ruc.dk Tel: +613-8344-9155 Tel: +45 46 74 38 32 Fax: +613-9348-1184 Programme Committee: Jose Alferes, Universidade Nova de Lisboa Francisco Bueno, Universidad Polytecnica de Madrid Henning Christiansen, Roskilde University Sandro Etalle, University of Twente Francois Fages, INRIA Maurizio Gabbrielli, University of Bologna Maria Garcia de la Banda, Monash University Michael Gelfond, Texas Tech University Gopal Gupta, UT Dallas Katsumi Inoue, Kobe University Joxan Jaffar, National University of Singapore Gerda Janssens, K.U.Leuven Bharat Jayaraman, State University of New York at Buffalo Michael Leuschel, University of Southampton Michael Maher, Loyola University Chicago Dale Miller, The Pennsylvania State University Ulf Nilsson, Linkoping University Francesca Rossi, Universita' di Padova Konstantinos Sagonas, Uppsala University Christian Schulte, Universitat des Saarlandes Harald Sondergaard, University of Melbourne Francesca Toni, Imperial College London Miroslaw Truszczynski, University of Kentucky Pascal Van Hentenryck, Brown University David S. Warren, State University of New York at Stony Brook ------- End of Blind-Carbon-Copy

1 0

question about kinds
by Hal Daume III 20 Jan '02

20 Jan '02

I have the following definition: > class Traversable d where > traverse :: d a -> (Maybe a, [d a]) And the standard binary tree data type: > data Tree a = Branch (Tree a) (Tree a) > | Leaf a I can define both Tree and [] to be instances of Traversable: > instance Traversable Tree where > traverse (Leaf a) = (Just a, []) > traverse (Branch t1 t2) = (Nothing, [t1,t2]) > instance Traversable [] where > traverse [] = (Nothing, []) > traverse (x:xs) = (Just x, [xs]) Now, I want to say that if some data type 'd' is Traversable and another data type 'e' is Traversable, then the "combined data type" is Traversable. That is, for example, I want to say that a Tree of Lists is traversable, or that a List of Trees, or a List of Lists is traversable. But I can say neither: > instance Traversable (Tree []) where ... or: > instance (Traversable a, Traversable b) => Traversable (a b) where .. Because of the obvious kind errors. What I suppose I need is some sort of lambda expansion over kinds, so I could say: > instance (Traversable a, Traversable b) => Traversable (\x -> a b x) or something like that. Obviously this doesn't exist. How can I get around this? - Hal -- Hal Daume III "Computer science is no more about computers | hdaume(a)isi.edu than astronomy is about telescopes." -Dijkstra | www.isi.edu/~hdaume

2 3

Re: question about kinds
by Ashley Yakeley 19 Jan '02

19 Jan '02

At 2002-01-18 13:19, Hal Daume III wrote: >In theory, I could write: > >class Traversable d a where > traverse :: d a -> (Maybe a, [d a]) > >instance Traversable Tree a where > traverse (Leaf a) = (Just a, []) > traverse (Branch t1 t2) = (Nothing, [t1,t2]) > >instance Traversable [] a where > traverse [] = (Nothing, []) > traverse (x:xs) = (Just x, [xs]) > >instance (Traversable d (e a), Traversable e a) => Traversable d (e >a) where > traverse = ... > >but then both ghc and hugs complain about overlapping instances, even with >-fallow-overlapping-instances/-fallow-undecidable-instances in ghc and +o >in hugs. Try this: class Traversable t a where -- note no fundep traverse :: t -> (Maybe a, [t]) instance Traversable (Tree a) a where traverse (Leaf a) = (Just a, []) traverse (Branch t1 t2) = (Nothing, [t1,t2]) instance Traversable [a] a where traverse [] = (Nothing, []) traverse (x:xs) = (Just x, [xs]) instance (Traversable (d (e a)) (e a), Traversable (e a) a) => Traversable (d (e a)) a where traverse = ... ...or even, more generally... instance (Traversable a b, Traversable b c) => Traversable a c where traverse = ... -- Ashley Yakeley, Seattle WA

1 0

Re: question about kinds
by Ashley Yakeley 19 Jan '02

19 Jan '02

At 2002-01-18 13:10, Hal Daume III wrote: >Now, I want to say that if some data type 'd' is Traversable and another >data type 'e' is Traversable, then the "combined data type" is >Traversable. That is, for example, I want to say that a Tree of Lists is >traversable, or that a List of Trees, or a List of Lists is traversable. If the Tree type constructor is Traversable, then it's Traversable no matter what it's applied to. You've provided a instance for traversing "Trees of anything", it's going to overlap with any instance for "Trees of Lists". -- Ashley Yakeley, Seattle WA

1 0

Re: dependent type query
by Ashley Yakeley 19 Jan '02

19 Jan '02

At 2002-01-18 12:01, Hal Daume III wrote: >class MyClass f a | a -> f, f -> a where ... > >I want to write a data type: > >data (MyClass f a) => MyData f = MyData (f, a) > >But this doesn't seem to be valid because "a" isn't in scope. Class constraints in 'data' declarations only affect the type of constructor functions. In other words, data (Context) => MyData a b c = MkMyData whatever is equivalent to data MyData a b c = MkMyData whatever except for the type of MkMyData, which will be MkMyData :: (Context) => whatever -> MyData a b c instead of MkMyData :: whatever -> MyData a b c > But a is uniquely determined by f. What gives? What you want is data MyData f = forall a. (MyClass f a) => MyData (f, a) >Furthermore, I can't even write: > >data (MyClass f a) => MyData2 f = MyData2 f Well now this is odd. I had no trouble simulating it with this: data MyData2 f = MyData2 f mkMyData2 :: (MyClass f a) => f -> MyData2 f mkMyData2 = MyData2 Looks like extended Haskell is being excessively restricted. Comments? -- Ashley Yakeley, Seattle WA

1 0

dependent type query
by Hal Daume III 18 Jan '02

18 Jan '02

Suppose I have the following definition: class MyClass f a | a -> f, f -> a where ... I want to write a data type: data (MyClass f a) => MyData f = MyData (f, a) But this doesn't seem to be valid because "a" isn't in scope. But a is uniquely determined by f. What gives? Furthermore, I can't even write: data (MyClass f a) => MyData2 f = MyData2 f This makes no sense to me whatsoever. This problem can the thwarted by moving the contraint to only the function definitions, but for MyData itself, it's really a hassle. Please could someone explain what's going on? -- Hal Daume III "Computer science is no more about computers | hdaume(a)isi.edu than astronomy is about telescopes." -Dijkstra | www.isi.edu/~hdaume

1 0

RE: (no subject)
by Simon Peyton-Jones 18 Jan '02

18 Jan '02

Perfectly sensible suggestion, but too late for the H98 Report, I'm afraid. It's time to freeze it. Simon | -----Original Message----- | From: D.Wakeling(a)exeter.ac.uk [mailto:D.Wakeling@exeter.ac.uk] | Sent: 08 January 2002 10:02 | To: haskell(a)haskell.org | Subject: (no subject) | | | | Hello, | | Is there any reason why the "gap" in a string literal is defined as | | gap -> \whitechar {whitechar}\ | | (in Section 2.6 of the report), rather than | | gap -> \whitespace\ | | ? The second, more liberal, definition of "gap" seems more | natural. Moreover, it would allow comments in the gaps, which | I would find useful. For example, | | classHeader :: String | classHeader = "\xca\xfe\xba\xbe\ -- magic number | \\x00\x03\ -- minor version | \\x00\x2d" -- major version | | David | -------------------------------------------------------------- | ----------------- | Dr. David Wakeling, School of Engineering and Computer Science | University of Exeter, Exeter, Devon, EX4 4PT. | unofficial web page: http://www.dcs.ex.ac.uk/~david | -------------------------------------------------------------- | ----------------- | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell |

1 0