December 2008 - Haskell-Cafe

FunDeps vs. Associated Types
by Sebastian Fischer 05 Dec '08

05 Dec '08

Dear Haskellers, I have a question regarding the correspondence between functional dependencies and associated types. > {-# LANGUAGE TypeFamilies, > FlexibleInstances, > MultiParamTypeClasses, > FunctionalDependencies > #-} With associated types, we can define a (useless[^1]) type class > class Useless a > where > type T a > useless :: a -> T a and instances > instance Useless () > where > type T () = () > useless = id > > instance Useless a => Useless (() -> a) > where > type T (() -> a) = T a > useless f = useless (f ()) Now we can compute `()` in many different ways: useless () useless (\()->()) ... I thought I could express the same with a multi-parameter type class and a functional dependency: > class UselessFD a b | a -> b > where > uselessFD :: a -> b But the corresponding instances > instance UselessFD () () > where > uselessFD = id > > instance UselessFD a b => UselessFD (() -> a) b > where > uselessFD f = uselessFD (f ()) are not accepted (at least by ghc-6.10.1) without allowing undecidable instances: useless.lhs:50:2: Illegal instance declaration for `UselessFD (() -> a) b' (the Coverage Condition fails for one of the functional dependencies; Use -XUndecidableInstances to permit this) In the instance declaration for `UselessFD (() -> a) b' Is there a simple explanation for this? Cheers, Sebastian [^1]: Originally, I was implementing hidden generation of unique identifiers. So instead of `useless :: (() -> () -> ... -> ()) -> ()` I got something like `withUnique :: (ID -> ... -> ID -> a) -> a`.

2 1

two type-level programming questions
by Nicolas Frisby 05 Dec '08

05 Dec '08

1) Type families, associated types, synonyms... can anything replace the use of TypeCast for explicit instance selection? Section 2, bullet 4 of http://www.haskell.org/haskellwiki/GHC/AdvancedOverlap indicates a negative response. Any other ideas? 2) Any progress/options for kind polymorphism in instances? Thanks for your time.

3 3

manipulating predicate formulae
by Ganesh Sittampalam 05 Dec '08

05 Dec '08

Hi, Are there any Haskell libraries around for manipulating predicate formulae? I had a look on hackage but couldn't spot anything. I am generating complex expressions that I'd like some programmatic help in simplifying. Cheers, Ganesh

4 7

Hat cannot use with ghc 6.10?
by Magicloud 05 Dec '08

05 Dec '08

Hi, I want to do some tracing, and I thought hat could help. Well hmake and hat both could not been made with ghc 6.10....

1 0

Finall Call For Papers (DSL WC)
by Emir Pasalic 05 Dec '08

05 Dec '08

*** IFIP Working Conference on Domain Specific Languages (DSL WC) *** July 15-17, 2009, Oxford http://www.hope.cs.rice.edu/twiki/bin/view/WG211/DSLWC * Call for Papers Domain-specific languages are emerging as a fundamental component of software engineering practice. DSLs are often introduced when new domains such as web-scripting or markup come into existence, but it is also common to see DSLs being introduced and adopted for traditional domains such as parsing and data description. Developing software using DSLs has many benefits. DSLs are often designed based on existing notations that are already in use by experts in a given domain. As such, successful DSLs often reduce or eliminate the effort needed to transform the concept or innovation produced by the domain expert into an executable artifact or even a deliverable software product. DSL implementations can capture and mechanize a significant portion of the repetitive and mechanical tasks that a domain expert traditionally needed to perform in order to produce an executable. DSLs can in many cases capture and make widely available special expertise that only top specialists in a given domain might have. By capturing expert knowledge and reducing repetitive tasks, DSLs often also lead to software that is significantly more portable, more reliable and more understandable than it would otherwise be. DSLs can be viewed as having a dual role to general-purpose languages: whereas general purpose languages try to do everything as well as possible, DSLs are designed to find a domain where they can solve some class of problems -- no matter how small -- in the best possible way. Widely known examples of DSLs include Matlab, Verilog, SQL, LINQ, JavaScript, PERL, HTML, Open GL, Tcl/Tk, Macromedia Director, Mathematica/Maple, AutoLisp/AutoCAD, XSLT, RPM, Make, lex/yacc, LaTeX, PostScript, Excel, among many others. But while these tools have been widely successful, they still fall short of realizing the full idea behind them. The goal of this conference is to explore the extent to which incorporating modern principles of language design and software engineering can benefit existing and future domain-specific languages. The ultimate goal of using DSLs is to improve programmer productivity and software quality. Often, this is achieved by reducing the cost of initial software development as well as maintenance costs. These improvements - programs being easier to write and maintain - materialize as a result of domain-specific guarantees, analyses, testing techniques, verification techniques, and optimizations. * Paper Criteria Papers are sought addressing the research problems, fundamental principles, and practical techniques of DSLs, including but not limited to: - Foundations, including semantics, formal methods, type theory, and complexity theory - Language design, ranging from concrete syntax to semantic and typing issues - Software engineering, including domain analysis, software design, and round-trip engineering - Software processes, including metrics for software and language evaluation - Implementation techniques, including parsing, compiling, and program generation - Program analysis and automated transformation - Reverse engineering, re-engineering, design discovery, automated refactoring - Hardware/software codesign - Programming environments, including visual languages, debuggers, and testing infrastructure - Teaching DSLs and the use of DSLs in teaching Case studies, including engineering, bioinformatics, hardware specification languages, parallel computing languages, real-time and embedded systems, and networked and distributed domains Papers will be judged on the depth of their insight and the extent to which they translate specific experience into general lessons for domain-specific language designers and implementers, and software engineers. Papers can range from the practical to the theoretical; where appropriate, they should refer to actual languages, tools, and techniques, provide pointers to full definitions and implementations, and include empirical data on results. * Important Dates - July 23rd, 2008: First Call for Papers - November 12th, 2008: Final Call for Papers - December 14th, 2008: Abstract submission due. - December 21st, 2008: Paper submission deadline. - February 23rd, 2009: Author notification of decisions - March 22nd, 2009: Camera ready manuscripts due * Instructions for Authors Proceedings will be published in the Springer LNCS series. Submissions and final manuscripts are to follow the LNCS stylesheet formatting guidelines, and are not to exceed 25 pages. Please submit your manuscripts online using the EasyChair conference management system. * Program Committee Jon Bentley, Avayalabs Martin Erwig, Oregon State University Jeff Gray, University of Alabama at Birmingham Robert Grimm, New York University Jim Grundy, Intel Strategic CAD Labs Tom Henzinger, EPFL Sam Kamin, UIUC Dick Kieburtz, Portland State University Ralf Lämmel, University of Koblenz Julia Lawall, University of Copenhagen Benjamin Pierce, University of Pennsylvania Vivek Sarkar, Rice University Jeremy Siek, University of Colorado at Boulder José Nuno Oliveira, University of Minho Doaitse Swierstra, Utrecht University Walid Taha (Chair), Rice University Eelco Visser, Delft University William Waite, University of Colorado at Boulder Stephanie Weirich, University of Pennsylvania * Organizers General Chair: Jeremy Gibbons, Oxford University Publicity Chair: Emir Pasalic, LogicBlox

1 0

class method name scope
by Jason Dusek 05 Dec '08

05 Dec '08

What proposals are out there to address the issue of scoping class methods? I always feel I must be careful, when exposing a class definition that I want clients to be able to extend, that I mustn't step on the namespace with semantically appropriate but overly general names (e.g. 'run'). It'd be nice if class method names were module scoped and could be qualified. -- _jsn

2 4

Re: [Haskell-cafe] detecting socket closure in haskell
by Donn Cave 05 Dec '08

05 Dec '08

Quoth "Tim Docker" <timd(a)macquarie.com.au>: | Hence I seem to need a means of detecting that a socket has been closed | remotely, without actually reading from it. . Does anyone know how to do | this? One reference I've found is this: | | http://stefan.buettcher.org/cs/conn_closed.html | | Apparently recv() with appropriate flags can detect this - though I'm | not convinced that the code as shown on that page doesn't busy wait when | there is unread data from the client. Right - poll() will still find data there, after the "peek", so you'll be running that loop as fast as you can make the 2 system calls, and what I think might be worse, it should keep reading the same 32 bytes at the head of the buffer ... and never notice the connection close. Well, I haven't tried it, but this solution doesn't make sense to me. | Any tips or pointers? Perhaps what I'm trying to do is not currently | possible in haskell, without using the FFI (which would be ok). Or | perhaps it's not possible in linux at all. I don't know of a way to do it in C. I think it's very likely that the kernel finds out only after all the data has been delivered through the network interface, so by the time you can even in principle know that the connection closed, all the data will either have been processed by the client or reside in memory somewhere on the computer. I guess you might try your buffering thread, and if you're really worried about undue amounts of buffered data, set some maximum size, where you either stop trying to notice end-of-file, or switch to another storage medium (e.g., disk file.) Donn

1 0

detecting socket closure in haskell
by Tim Docker 04 Dec '08

04 Dec '08

This is a haskell + networking question... I have a multi threaded haskell server, which accepts client connections, processes their requests, and returns results. It currently works as desired, except where a client drops a connection whilst the server is processing a request. In this circumstance, the server currently doesn't notice that the client has gone until it finishes the processing and attempt to read the next request. I'd like to change it so that the request is aborted as soon as the client disconnects. One way of doing this would would be to maintain a separate thread that is always reading from the client, and buffers until the main thread needs the information. Whilst this would detect the remote close, it also would potentially consume large amounts of memory to maintain this buffer. Hence I seem to need a means of detecting that a socket has been closed remotely, without actually reading from it. . Does anyone know how to do this? One reference I've found is this: http://stefan.buettcher.org/cs/conn_closed.html Apparently recv() with appropriate flags can detect this - though I'm not convinced that the code as shown on that page doesn't busy wait when there is unread data from the client. Any tips or pointers? Perhaps what I'm trying to do is not currently possible in haskell, without using the FFI (which would be ok). Or perhaps it's not possible in linux at all. Thanks, Tim

2 1

ANN: "Real World Haskell", now shipping
by Bryan O'Sullivan 04 Dec '08

04 Dec '08

Good evening - John Goerzen, Don Stewart and I are delighted to announce the availability of our book, "Real World Haskell". It is 710 pages long, and published by O'Reilly Media. This is the first book to comprehensively cover modern Haskell programming. From an introduction to functional programming, it focuses on teaching through many worked examples. We discuss the "awkward squad" of I/O, concurrency, and exceptions. We cover network programming, databases, and system hacking. We motivate and work with monoids, applicative functors, monads, and monad transformers. We show you how to debug code, and how to ship well-tested software. Better yet, the book is available under a Creative Commons license, so you can read as much of it as you please before you buy: http://book.realworldhaskell.org/ We developed this book with the enthusiastic and voluble support of the Haskell community, and we are proud to share our work in a fashion that will help newcomers to our field. And best of all, if you order now (at least in North America), you can have a copy of the book in your hands in a matter of days. Thank you from all of us to our friends in the Haskell world who have been so generous with their feedback and kind words!

17 34

gmp license == no commercial/closed source haskell software ??
by Lionel Barret De Nazaris 04 Dec '08

04 Dec '08

I've just discovered the GMP license problem. (see http://hackage.haskell.org/trac/ghc/wiki/ReplacingGMPNotes) From my understanding, the gmp is GPL, GHC statically links it on windows. As a consequence, any program compiled using GHC must be distributed under a GPL compatible license. In the threads I've read, some workarounds are proposed on linux and OsX, but I didn't see anything on windows. Is the situation still the same ? Does anybody know a work-around to make a closed-source programm in haskell without violating the GPL ? ( Note this could block Haskell in our company. ) -- Best Regards, lionel Barret de Nazaris, ========================= Gamr7 : Cities for Games http://www.gamr7.com

6 5