November 2001 - Haskell

Re: How do I write the type of 'inverse'?
by John Hughes 16 Nov '01

16 Nov '01

I'll try explaining via example: I am trying to define a `Pairable' class, which is an abstraction of all kinds of pairs: (a,b), Pair a b, etc. > class Ord p => Pairable p where > is_pair :: p -> Bool > pair_fst :: Ord a => p -> a -- Precondition: is_pair > pair_snd :: Ord b => p -> b -- Precondition: is_pair > make_pair :: (Ord a, Ord b) => a -> b -> p > instance (Ord a, Ord b) => Pairable (a,b) where > is_pair (a,b) = True > pair_fst (a,b) = a > pair_snd (a,b) = b > make_pair a b = (a,b) Funny: I get an error from just this instance declaration! The problem is that the class definition means that pair_fst is *polymorphic* in a, that is, it should be able to return *any* type in class Ord. Of course, your instance can't: it can only return the type of the first component. Thus the instance is not general enough. The solution (using hugs -98 or ghc -fglasgow-exts, because this is an extension to Haskell 98) is to use a multi-parameter class: > class (Ord p, Ord a, Ord b) => Pairable p a b where > is_pair :: p -> Bool > pair_fst :: p -> a -- Precondition: is_pair > pair_snd :: p -> b -- Precondition: is_pair > make_pair :: a -> b -> p > instance Pairable (a,b) a b where > is_pair (a,b) = True > pair_fst (a,b) = a > pair_snd (a,b) = b > make_pair a b = (a,b) This is what you were after here: It seems like I want to pass the component types into the class definition somehow: class Pairable (p a b) Now the instances of pair_fst and pair_snd are for a *particular* component type, and all is fine. This works, but tends to lead to many ambiguities, since the type-checker cannot know that there isn't an instance instance Pairable (a,b) b a where ... pair_fst (a,b) = b ... also. That means it cannot infer, when it sees pair_fst (a,b), that the result is of the same type as a. You can avoid the ambiguities too, by adding a *functional dependency* to the class definition. In this case, the type of the "pair" presumably determines the type of each component, so you could write > class (Ord p, Ord a, Ord b) => Pairable p a b | p->a, p->b where > is_pair :: p -> Bool > pair_fst :: p -> a -- Precondition: is_pair > pair_snd :: p -> b -- Precondition: is_pair > make_pair :: a -> b -> p to declare that for each type p, there can only be one type a and one type b. This permits the type-checker to infer the result type of pair_fst and pair_snd from the argument type. Of course, if you actually *want* to declare pairs with reversed components as "pair-like" also, then you can't do this: you have to put up with the ambiguities (and add types by hand to resolve them) because the ambiguities are really there. John

1 0

RE: GHC installation
by Simon Peyton-Jones 16 Nov '01

16 Nov '01

You don't say which version of GHC... GHC 5.02.1 (which I advise you to use) installs in c:/ghc/ghc-5.02.1 The installer doesn't give you a way to change this. (It should, but we havn't found out how yet.) But once you have put it there you can move it if you want. (Best to mail glasgow-haskell-users rather than the entire Haskell mailing list with GHC-related questions.) Simon | -----Original Message----- | From: Jerzy Karczmarczuk [mailto:karczma@info.unicaen.fr] | Sent: 16 November 2001 11:56 | Cc: haskell(a)haskell.org | Subject: GHC installation | | | I might be dead wrong, in that case I apologize... | | I just took the Windows installer and tried to put GHC etc. | on my laptop. I suspect that the installer absolutely wants | to put the stuff in Program Files and doesn't give the user | the opportunity to install it on another disk. Anyway, I have | plenty of space on another partition, but the installer | complains that it lacks space. | | Any suggestions, please? | | | Jerzy Karczmarczuk | Caen, France | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell |

1 0

RE: how strict is Array.array?
by Simon Peyton-Jones 16 Nov '01

16 Nov '01

Well, it's clearly a design choice. But in general laziness does not come for free. The choice made by the Haskell designers was that the 'array' function can consume and explore all its input before producing any result. It just give more scope to the implementor. Nothing too deep. Simon | -----Original Message----- | From: Johannes Waldmann [mailto:joe@isun.informatik.uni-leipzig.de] | Sent: 12 November 2001 20:29 | To: haskell(a)haskell.org | Subject: how strict is Array.array? | | | I was surprised to note that | | bounds $ array (0,1) undefined | | is undefined ( rather than (0,1) ) - why is this? | | just curious, | -- | -- Johannes Waldmann ---- | http://www.informatik.uni-leipzig.de/~joe/ -- | -- | joe(a)informatik.uni-leipzig.de -- phone/fax (+49) 341 9732 204/252 -- | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell |

1 0

RE: What are ZMZN and ->Z1T in a ghc space profile?
by Julian Seward (Intl Vendor) 16 Nov '01

16 Nov '01

I suspect you're both right. I seem to remember that it used to be the way Keith says, but was relatively recently changed to be the way Robert says. J | > > But, to answer your emmediate question: | > > | > > ZMZM = [] - The list Nil constructor | > > Z1T = ( ) - The 1-tuple constructor | > | > IIRC, the "1" is the number of commas, so Z1T is the *pair* | > constructor. | | The comments in OccName.lhs say: | | [copying direct from the file] | | Before After | -------------------------- | Trak Trak | foo_wib foozuwib | > zg | >1 zg1 | foo# foozh | foo## foozhzh | foo##1 foozhzh1 | fooZ fooZZ | :+ ZCzp | () Z0T 0-tuple | (,,,,) Z5T 5-tuple | (# #) Z1H unboxed 1-tuple (note the space) | (#,,,,#) Z5H unboxed 5-tuple | (NB: There is no Z1T nor Z0H.) | | Which, interestingly, says that there is no Z1T, despite | sengan having | supposedly found one. | | However it also suggests that, if Z1T did exist, then it | would be a 1-tuple. | | | The code itself backs this up and suggests that the number is | the number of | commas + 1: | | maybe_tuple "()" = Just("Z0T") | maybe_tuple ('(' : cs) = case count_commas (0::Int) cs of | (n, ')' : cs) -> Just ('Z' : | shows (n+1) "T") | other -> Nothing | maybe_tuple other = Nothing | | | Just in case anyone wasn't confused yet :-) | | | | -Rob | | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell |

1 0

What are ZMZN and ->Z1T in a ghc space profile?
by senganb＠ia.nsc.com 16 Nov '01

16 Nov '01

I searched on google, but had no luck. My program is an implementation of http://www-aig.jpl.nasa.gov/public/home/gat/lisp-study.html in haskell, and the dictionary read in is rather large (900Kb) so I assume most of the space should be [[Char]] Sengan

3 3

Re: Student Programming Projects
by Yoann Padioleau 16 Nov '01

16 Nov '01

Johannes Waldmann <joe(a)isun.informatik.uni-leipzig.de> writes: > > Next Semester, I am supposed to teach a short course in Haskell. > > Can anyone recommend interesting programming projects which can > > be completed in about a month? Thank you very much. > > apart from "pure" programming exercise > (see Haskell textbooks http://haskell.org/bookshelf/) > here are a few suggestions in the area of experimental mathematics. > also, check with your introductory discrete mathematics course (or equivalent). > a lot of combinatorial problems on finite structures > can be illustrated by finding (counter-)examples by computer. > > > say Ramsey colorings, Monochromatic arithmetic subsequences > (van der Waerden's theorem), words avoiding patterns, and so on. > http://www-igm.univ-mlv.fr/~berstel/Lothaire/ChapitresACW/C3.ps > > next, perhaps some game theory? > nim (and variants), other take-and-break games (octal games). > you can safely ignore all of the Sprague-Grundy theory, > and just define the game, and ask for some method > to determine quickly who wins. > > or graph theory: finding hamiltonian paths > (knights tours on a x b chessboards, for instance) > > if the students alreay took some computability theory, > then a nice topic is Posts Correspondence Problem, see > http://www.cs.ualberta.ca/~zhao/PCP/intro.htm > http://www.informatik.uni-leipzig.de/~pcp/pcpcontest_en.html > > also string rewriting: finding/avoiding loops. > this (open!) problem is easily stated: http://www.lri.fr/~rtaloop/95.html Wow, what some nice exercise, are you really serious ? do you think computer science student enjoy that kind of programs (except for the AI-related stuff)? why not make them programming something they would like to use !! why not a tetris/pong/reversi/any_video_game, see for example the book from Paul hudak (multimedia ... in haskell). this man understand what makes student happy : graphism,video games,compiler,operating systems, AI games and not this shit of mathematics/computer science theorist problem. This is the kind of things that make student think that functionnal programming is for doing useless things, and that C is cool cos you can do graphism, program your own operating systems, ... ICFP 2000 ask for an interesting problem: a raytracer, a thing that many student like cos you can then play with it and describe your own scene, and that contest show that ocaml/haskell were really cool for doing graphism-related work. > > > hope this helps. I'm happy to provide more detail. -- Yoann Padioleau, INSA de Rennes, France, http://www.irisa.fr/prive/padiolea Opinions expressed here are only mine. Je n'écris qu'à titre personnel. **____ Get Free. Be Smart. Simply use Linux and Free Software. ____**

2 1

CFP: EuroPar 20002
by Kevin Hammond 16 Nov '01

16 Nov '01

[Apologies for any duplication. Please can you distribute to those who might be interested... I am global chair of topic 10, Parallel Programming: Models, Methods and Programming Languages, and would welcome submissions from the Haskell community. Kevin] ***** Euro-Par 2002 - http://europar.upb.de ****** Paderborn, Germany August 27-30, 2002 First Call for Papers Euro-Par is well established as the premier European conference on all aspects of parallel computing. Previous events were held in Stockholm (1995), Lyon (1996), Passau (1997), Southampton (1998), Toulouse (1999), Munich (2000), and Manchester (2001). The conference normally attracts more than 300 participants coming from universities, research centres, and industry. The major themes of the conference can be divided into the broad categories of hardware, software, algorithms and applications. In common with previous years, Euro-Par 2002 will be organised as a number of parallel sessions on the following topics for which papers are solicited: 1. Support Tools and Environments 2. Performance Evaluation, Analysis and Optimization 3. Scheduling and Load Balancing 4. Compilers for High Performance (compilation and parallelization techniques) 5. Parallel and Distributed Databases, Data-Mining and Knowledge Discovery 6. Complexity Theory and Algorithms 7. Applications of High-Performance Computers 8. Parallel Computer Architecture and Instruction Level Parallelism 9. Distributed Systems and Algorithms 10. Parallel Programming: Models, Methods and Programming Languages 11. Numerical Algorithms 12. Routing and Communication in Interconnection Networks 13. Architectures and Algorithms for Multimedia Applications 14. Meta- and Grid-Computing 15. Discrete Optimization 16. Mobile Computing, Mobile Networks Further details for each topic can be found at: http://europar.upb.de/ Authors are requested to use the electronic form on the web site to submit their paper to the topic they judge most appropriate. All accepted papers will be available at the conference in the proceedings published by Springer-Verlag in the LNCS series. The conference will also feature invited talks and tutorials. Further information will appear on the website as it becomes available. -------------------------------------------------------------- Key Dates: February 8th, 2002 Final date for submissions April 30th, 2002 Acceptance notified May 31st, 2002 Deadline for final papers June 30th, 2002 Author registration deadline June 30th, 2002 Early registration deadline August 27th-30th, 2002 Euro-Par 2002 in Paderborn Apologies for multiple copies of this call for papers! If you would like to be removed from this mailing list or have received multiple copies and you would like some address(es) to be removed please email europar(a)upb.de

1 0

Contact Information
by Brian M. Ratzker 16 Nov '01

16 Nov '01

Could you please help me out? Who could I talk to in your company regarding marketing your web site? Would appreciate any help you can offer. Thanks Brian ====================================================== QUALIFIED TARGETED LEADS GUARANTEED!!! SIGN-UP as a myGeek Seller and deposit $20.00 between November 1, 2001 and November 30th, 2001 and we guarantee that you will receive Qualified Targeted Leads. If you're not satisfied with our service for any reason, within the first 30 days, we will REFUND your $20.00 Just click on the following link to learn more about myGeek.com: http://www.mygeek.com/login.jsp?MNO=98008&AID=1&PID=1 ===================================================== Thanks again, Brian Ratzker myGeek.com, Inc. 480.421.2200 ext 257

1 0

PhD Studentship (Please Forward)
by Kevin Hammond 15 Nov '01

15 Nov '01

Applications are invited for a PhD student to work on resource modelling for Hume, a concurrent bounded resource functional language aimed at embedded/safety critical systems. The student must be UK- or EC-resident, and possess a good undergraduate degree in Computer Science. The project will pay both fees and a maintenance grant at the standard rate. The studentship can be taken up either in the near future or later, to suit individual circumstances (e.g. at the start of the next academic year, September 2002). More information on the Hume language can be found at: http://www-fp.dcs.st-and.ac.uk/hume Please address informal enquiries to kh(a)dcs.st-and.ac.uk. The project is funded by the UK's Engineering and Physical Sciences Research Council (EPSRC). The prototype implementation is written in Haskell. We hope to make it available to interested parties in due course! Kevin

1 0

ANNOUNCE: linkchk-0.02 released
by Jens-Ulrik Petersen 14 Nov '01

14 Nov '01

Happy to announce linkchk version 0.02 is out. >From the NEWS file: * width of window is now fixed * new options to set: ** window width ** host to be pinged ** interface * checks ifconfig to see if interface is up * use implicit parameters for command-line options The tarball and i386 rpm are now available on: http://www.01.246.ne.jp/~juhp/haskell/linkchk/ linkchk is also on freshmeat.net. Enjoy, Jens

1 0