Haskell Weekly News: September 18, 2006
Welcome to issue 41 of HWN, a weekly newsletter covering developments in the Haskell community. Each week, new editions are posted to the Haskell mailing list as well as to the Haskell Sequence and Planet Haskell. RSS is also available, and headlines appear on haskell.org.
The 2006 Haskell Workshop was held today in Portland, Oregon. Thanks to Edward Kmett for a report on the event.
Haskell98 Termination Analyser . Stephan Swidersk announced the integration of an automatic Haskell98 termination analyzer in the termination tool AProVE. The tool accepts full Haskell as specified in the Haskell 98 Report and is available through our web interface. More
Free theorems . Janis Voigtlaender announced that Sascha Boehme has done a project to implement the Reynolds/Wadler algorithm generating theorems from polymorphic types, plus simplifications and postprocessings for such free theorems. More info
Haddock/GHC SoC . David Waern announced a short status report of the "Port Haddock to use GHC" Summer of Code project. The GHC modifications, are finished and will be included in the GHC head repository soon.
SOE implementation based on Gtk2Hs . Duncan Coutts Due to popular demand the new SOE implementation based on Gtk2Hs is now available. The rendering quality is better than the original HGL version. Here's a side-by-side comparison
SmallCheck . Colin Runciman released a prototype tool that is similar in spirit, and in some of its workings, to QuickCheck. SmallCheck is, though, based on exhaustive testing in a bounded space of test values. More info
Frisby: composable, linear time parser for arbitrary PEG grammers . John Meacham released Frisby, an implementation of the 'packrat' parsing algorithm, which parse PEG grammars and have a number of very useful qualities, they are a generalization of regexes in a sense that can parse everything in LL(k), LR(k), and more, including things that require unlimited lookahead, all in guaranteed linear time. More information
HaskellNet . Jun Mukai published a status report on the state of his SoC project, HaskellNet
GHC's new support engineer . Simon Marlow announced that GHC now has a full-time support engineer, Ian Lynagh (aka Igloo on IRC). He'll be helping with all aspects of GHC, especially release management, bug diagnosis and tracking, documentation, packaging, and supporting other GHC hackers. Welcome Ian!
Haskell'This section covers the Haskell' standardisation process.
- Writing / status teams - call for volunteers
- do and if then else
- map and fmap
- All monads are functors
The Haskell Workshop . Here is a summary of the talks presented today at the Haskell Workshop
RepLib: A Library for Derivable Type Classes . Stephanie Weirich presented a neat library for generic traversals. It covered just about everything you'd need and seemed pretty painless to use. From the abstract:
Some type class instances can be automatically derived from the structure of types. As a result, the Haskell language includes the "deriving" mechanism to automatic generates such instances for a small number of built-in type classes. In this paper, we present RepLib, a GHC library that enables a similar mechanism for arbitrary type classes.
A Generic Recursion Toolbox for Haskell (Or: Scrap Your Boilerplate Systematically) . Deling Ren and Martin Erwig talked about another generics library, with a neat core "mother traversal" that abstracted over different data structure traversals.
The Recursion Library for Haskell provides a rich set of generic traversal strategies to facilitate the flexible specification of generic term traversals. The underlying mechanism is the Scrap Your Boilerplate (SYB) approach. Most of the strategies that are used to implement recursion operators are taken from Stratego.
Strong Types for Relational Databases . Alexandra Silva and Joost Visser presented a relational database Using the type system to encode not only the types of fields it was joining, but a bit more about their meaning and role in the key.
Type-level programming is used to maintain the consistency between header and rows, and between functional dependencies and tables. Type-level predicates are provided to check whether tables and functional dependencies arein particular normal-forms, or not.
Polymorphic Variants in Haskell . Koji Kagawa presented a paper on polymorphic variants in Haskell, describing some interesting type system extensions.
In languages that support polymorphic variants, a single variant value can be passed to many contexts that accept different sets of constructors. Polymorphic variants are potentially useful for application domains such as interpreters, graphical user interface (GUI) libraries and database interfaces, where the number of necessary constructors cannot be determined in advance.
Extended Static Checking for Haskell . Dana Xu gave a great talk on ESC/Haskell, proving the partial correctness of programs with compile time checks for pre/postcondition violations, and using an external theorem prover to prove bits about arithmetic expressions. Very practical stuff. The extended static checking is beautiful in that it opens up a whole new arena in Haskell for machine checkable documentation.
Running the Manual: An Approach to High-Assurance Microkernel Development . Philip Derrin, Kevin Elphinstone, Gerwin Klein, David Cock and Manuel M.T. Chakravarty presented a microkernel paper. This was really good. Before I saw the talk I really questioned the judgement of prototyping a microkernel in Haskell. Now I want to go prototype one myself.
We propose a development methodology for designing and prototyping high assurance microkernels, and describe our application of it. The methodology is based on rapid prototyping and iterative refinement of the microkernel in a functional programming language.
Strongly Typed Memory Areas -- Programming Systems-Level Data Structures in a Functional Language . Iavor S. Diatchki and Mark P. Jones gave a presentation on how you could extend the type system to include an extra kind for physical memory regions and data representations in memory, with endianness, and the like. He implemented it in a kind of strict pseudo-Haskell, but the general notation was pretty clean and would be usable for system level programming in Haskell.
User-Level Transactional Programming in Haskell . Peter Thiemann wrote on generalizing the interfacefor STM to other kinds of database and user-interface transactions, where he gave up the ability to automatically retry blocking on a changeset of TVars, producing a kind of transaction that could actually fail and be manually retried.
An Extensible Dynamically-Typed Hierarchy of Exceptions . Simon Marlow presented an extensible exception hierarchy using type classes.
In this paper we address the lack of extensibility of the exception type in Haskell. We propose a lightweight solution involving the use of existential types and the Typeable class only, and show how our solution allows a fully extensible hierarchy of exception types to be declared, in which a single overloaded catch operator can be used to catch either specific exception types, or exceptions belonging to any subclass in the hierarchy.
Demo: Introducing the Haskell Equational Reasoning Assistant . Andy Gill came in and did a really neat, high energy, demo of an interactive equational reasoning/code refactoring tool. He actually loaded up the application and showed it refactoring live.
GenI: Natural Language Generation in Haskell . Eric Kow covered genI and gave a user's level overview of the features in Haskell that are useful in real code. The natural language generation stuff there was pretty nifty.
Demo: Typed Linear Algebra . Frederik Eaton covered a linear algebra library with ATLAS and LAPACK bindings in Haskell that uses type level enumerable values to provide its bounds. Neat stuff.
Demo: Interactive Debugging . David Himmelstrup unforutnately was unable to present his talk, due to illness. From the abstract:
Haskell's lazy nature is a large part of its beauty. Unfortunately, laziness complicates debugging in the "stop, examine, continue" sense, and hence tools such as Freja, Hat and Hood have risen. This paper describes the implementation of a "stop, examine, continue" debugger for Glasgow Haskell
Demo: Haskell Program Coverage tool, HPC . A demo was presented on how you can use HPC to interactively control QuickCheck, get it to run some cases, and check to see how much coverage you have had, then have it run more cases if thats not enough. The notion of looking at the difference in the coverage sets between a good run and a bad run to try to isolate bugs seemsreally useful.
Demo: On computable types . Oleg Kiselyov presented a "proof by example" that the Haskell type system is Turing complete, with a 5 line type system lambda calculus and a 6 line type level fibonacci.
Haskell' Status Report -- An Update on the Next Haskell Standard . Isaac Jones Isaac gave a talk on the state of Haskell'
Quotes of the Week
- Baughn : "I think I'm beginning to understand this language. I just defined: let 1 + 1 = 3"
- SamB : "Haskell is the only language I know with first-class support for imperative programming"
- David Amos : "The real reason for using Haskell is that the code comes out shorter, and is quicker to write, than in imperative languages .. What that means is, I can get much more done when I use Haskell"
- SyntaxNinja : "I recommend seeing if people have a major problem, then pouring concrete on them, and implementing what you want anyway"
- SamB : "I was just saying that encasing poeple in concrete is not the best solution to that problem"
- Sean Russell : "[On the wmii window manager mailing list] Hah! That's a laugh. Since when does 'it compiles' equate to 'it will run (correctly)'? We're talking about C, after all... not Haskell."
- Adaptec : "[In documentation for one of their chipsets] We have a disclaimer because there may be corner cases"
- audreyt : "Yes, I'm aware that I'm abusing GHC beyond its designed purpose"
- cjeris : "I have a little perl script that aliases gcc -freduced-suckage to ghc"
- dibblego : "Using logic in imperative programming language channels is generally considered trolling"
- edwardk : "lambdabot is the first irc bot I ever added to my buddy list"
- gmh33 : "Haskell makes so much more sense now and all I did was figure out the typing thing"
- musasabi : "reallyUnsafePointerEq# seems unsafe"
- sieni : "The advantage of Haskell is that it doesn't suck"
- therp : "Good morning. I think I have been dreaming of shift/reset continuations"
Contributing to HWN
To help create new editions of this newsletter, please see the
contributing information. Send
dons at cse.unsw.edu.au . The darcs repository is available at
darcs get http://www.cse.unsw.edu.au/~dons/code/hwn