- Haskell-Cafe - Haskell.org

Type family oddity
by Florian Weimer 05 Oct '08

05 Oct '08

I can't figure out why the following code doesn't compile with the October 2n GHC 6.10 beta (-XTypeFamilies -XFlexibleContexts) when the type declaration is not commented out. module T where type family RangeTrait c class InputRange r where remaining :: r -> Bool advance :: r -> r class (InputRange (RangeTrait s)) => Sequence s where erase :: RangeTrait s -> IO (RangeTrait s) -- erase_range :: (Sequence s) => RangeTrait s -> IO (RangeTrait s) erase_range r = if remaining r then do r' <- erase r erase_range r' else return r GHCi says the type is precisely as specified in the comment. However, when I activate the type declaration, GHC complains: T.hs:16:22: Couldn't match expected type `RangeTrait s' against inferred type `RangeTrait s2' In the first argument of `erase', namely `r' In a stmt of a 'do' expression: r' <- erase r In the expression: do r' <- erase r erase_range r' T.hs:17:22: Couldn't match expected type `RangeTrait s1' against inferred type `RangeTrait s2' In the first argument of `erase_range', namely `r'' In the expression: erase_range r' In the expression: do r' <- erase r erase_range r' Any suggestions? Is this a bug in GHC?

3 4

Total Functional Programming in Haskell
by Jason Dagit 05 Oct '08

05 Oct '08

Hello, I recently had someone point me to this thread on LtU: http://lambda-the-ultimate.org/node/2003 The main paper in the article is this one: http://www.jucs.org/jucs_10_7/total_functional_programming/jucs_10_07_0751_… It leaves me with several questions: 1) Are there are existing Haskell-ish implementations of the total functional paradigm? 2) Could we restructure Haskell so that it comes in 3 layers, total functional core, lazy pure partial functional middle, and IO outer layer? The idea with #2 is that similar to how we use the type system to separate the purely functional from monadic, and in particular IO actions, can we make the inner most part of Haskell total? Furthermore, could we do this in a way that makes it easy to intermingle the three layers the way we can mingle pure Haskell with IO actions? Maybe instead of using (->) as the function constructor for total functions we use a different symbol, say (|->), and the compiler knows to use the more specialized semantics on those definitions. I'm not sure how make this work for values that are total functional versus values that are just pure (partial) functional. I'm also interested in hearing about the optimizations that would be possible when all your expressions are strongly normalizing. Jason

9 14

Job Opportunity
by Michael Bott 05 Oct '08

05 Oct '08

Afternoon, I have been recommended to use this site from a couple of people I have spoken to. Sorry if this offends anyone but I have an amazing opportunity for a 2 Functional Programmers based in London. Functional Programming - (HASKELL / ERLANG / OCAML) - Financial This is an incredible opportunity for a FUNCTIONAL PROGRAMMER (Haskell / Erlang or Ocaml) to join an expanding London software house who specialise in Asset Management for top financial clients - they specialise in providing algorithmic execution solutions for hedge funds. My client are a leading provider in financial software, they are looking for a top Graduates or 1-2 years experienced Java / Functional Programmers to join the team and really make a difference to the company and work with the biggest clients in the UK. Desirable Skills - * Functional Programming - ( Haskell or Erlang or Ocaml to name 3). * Java * Degree Education - top University - 2:1 and above only. * Experience within the finance and/or banking sector would be highly advantageous. This is an URGENT requirement and my client are keen to get at least 2 exceptional developers into the team ASAP. You will join a relaxed team who work on exciting projects, and use the most up to date technologies........ To learn more about the role and the client please contact Mike at Lawrence Harvey. Please contact me if you are interested. Kind Regards, Mike Michael Bott Java Recruitment Specialist Lawrence Harvey Search & Selection Tel: +44 (0)208 875 4353 Fax: +44 (0)208 870 8116 Mobile: +44 (0)7966 554892 Email: m.bott(a)Lawrenceharvey.com <mailto:m.bott@Lawrenceharvey.com> Web: www.Lawrenceharvey.com <http://www.lawrenceharvey.com/> LinkedIn - http://www.linkedin.com/in/mikebott <http://www.linkedin.com/in/mikebott> The information transmitted contains confidential material and is intended only for the person or organisation to which it is addressed. Any review, retransmission, dissemination or other use of this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender and delete the material from any computer. No liability whatsoever is accepted by Lawrence Harvey Search & Selection in respect of the contents of this email.

2 1

Hackage Build Failures
by Cetin Sert 04 Oct '08

04 Oct '08

Hi, what is the best action to take if a package from hackage fails to build? Is there a recommended/established common way to deal with build failures/runtime bugs etc.? For example: cetin@linux-d312:~/Links/Elite/ac> cabal install wired Resolving dependencies... Downloading Wired-0.1.1... Configuring Wired-0.1.1... Preprocessing library Wired-0.1.1... Building Wired-0.1.1... [ 1 of 21] Compiling Data.Hardware.Internal ( Data/Hardware/Internal.hs, dist/build/Data/Hardware/Internal.o ) Data/Hardware/Internal.hs:198:15: Ambiguous occurrence `><' It could refer to either `Data.Hardware.Internal.><', defined at Data/Hardware/Internal.hs:129:4 or `Test.QuickCheck.><', imported from Test.QuickCheck at Data/Hardware/Internal.hs:11:0-21 cabal: Error: some packages failed to install: Wired-0.1.1 failed during the building phase. The exception was: exit: ExitFailure 1 Regards, CS

7 7

Restricted file reading monad
by George Pollard 04 Oct '08

04 Oct '08

Hello all, I'm currently working on a (toy) ID3 [1] tag reader, which made me think of a library which might be quite useful. The structure of an ID3 tag goes something like this: Header: - total size of tag - other header info A series of frames, each with: - total size of frame - other header info - frame data Since the ID3 tag as a whole has size information, I need to pass that into the frame-reading functions to ensure that I never read past the end of the total header. This exact same requirement is present within the frame-reading function itself; it must read the frame data without reading past the end of the frame. It would be nice to have some kind of monad or similar so that I could do: > readTag = do > size <- -- read size > -- read other header info > withLimit size $ -- read frames then the read-frames function would be able to do its work without having to worry about size-checking all the time, as this would be implicit. (And similarly with the frame-reading functions as well.) This could easily be extended to: > f x = readManyWithLimit size readerFunction where readManyWithLimit would return a list of results of applying the reader function as many times as possible within the size limit (and possibly returning some kind of error if the reader function 'bails early' and doesn't gobble right up to the end of the size limit). The thing is, I've not implemented a monad of any substantial size so I don't have a clue where to start. Calling it FileReader (for want of a better name), I guess it would almost be the same as a parsing monad, so the type would have a reference to the file Handle and the current index into the file, as well as the last point that it's allowed to read up to. So, umm... a rather open-ended question, but how would I go about this? [1]: http://www.id3.org/Developer_Information

4 3

Haddock Feature Proposal
by Jason Dagit 04 Oct '08

04 Oct '08

Hello, I was just thinking about how many devs put "TODO ...", "FIXME ...", or "HACK..." into comments. Do others think this could be nice optional information for Haddock to display? I think it could give people an idea for how mature, stable or clean an api is just by inspecting the above items when they are present. Thoughts? Thanks, Jason

2 1

outdated OpenSPARC webpage
by Wolfgang Jeltsch 04 Oct '08

04 Oct '08

Hello, the webpage under <http://haskell.org/opensparc/> ist outdated. There is still the line saying that students can now apply. Who is responsible for this page? Wouldn’t it be better to move this page to the wiki so that keeping it up to date is easier? Best wishes, Wolfgang

1 0

Haskell Weekly News: Issue 88 - October 4, 2008
by Brent Yorgey 04 Oct '08

04 Oct '08

--------------------------------------------------------------------------- Haskell Weekly News http://sequence.complete.org/hwn/20081004 Issue 88 - October 04, 2008 --------------------------------------------------------------------------- Welcome to issue 88 of HWN, a newsletter covering developments in the [1]Haskell community. An extra-short HWN this week, so you get an extra ten minutes to do something else during the time you would have normally spent reading the HWN! HWN-editor-approved activities for your ten minutes include eating cookies, playing [2]Fantastic Contraption, and writing a type checker in the type system while eating cookies. Announcements Arch Haskell News: Oct 4 2008. Don Stewart [3]sent out the newest Arch Haskell news --- now with 609 Haskell packages! Announcing OneTuple-0.1.0. John Dorsey [4]announced the release of the ground-breaking [5]OneTuple library, which adds the long neglected one-tuple to Haskell. It also turns out that the denizens of Haskell-cafe are completely unable to refrain from turning jokes into long-winded technical discussions about strictness and lifted types. Haskell protocol-buffers version 0.3.1. Chris Kuklewicz [6]announced the release of [7]protocol-buffers 0.3.1, with some functionality also split off into [8]protocol-buffers-descriptor and [9]hprotoc. The 'hprotoc' compiler for proto files to Haskell source code now takes a "-u" command-line option. When given, this turns on code generation to support loading, storing, and saving unknown fields. Discussion Stacking monads. Andrew Coppin began a long [10]discussion on monads, monad transformers, Applicative, MonadPlus, and related topics. planning for ghc-6.10.1 and hackage. Duncan Coutts began a [11]discussion on how to make the transition to GHC 6.10 as painless as possible, especially as it relates to the new base-4 package and Cabal. Proposal #2629: Data.List: Replace nub; add nubOrd, nubInt, nubWith. Bart Massey [12]proposed refactoring nub into a 'nubWith' function which can be specialized to efficient versions for Int and Ord. Blog noise [13]Haskell news from the [14]blogosphere. * Luke Palmer: [15]Laziness and the monad laws. Luke explains why making a Functor or Monad instance too lazy can be just as bad as making it too strict. * Paul R Brown: [16]The Haskell Platform and Lessons Learned Elsewhere. * Eric Kow (kowey): [17]darcs weekly news #6. * Mads Lindstroem: [18]Overlapping Instances in Haskell. * >>> Bill Six: [19]Dabbling with Haskell. Bill explores palindromic pangrams using Haskell. Quotes of the Week * ozy`: [on RWH] most authors are like "FUNCTIONAL PROGRAMMING IS FUNCTIONAL!!!" whereas these guys are more like "yeah but practical programming is practical. map wash_dish dishes" * BMeph: * wants an "Everything I know about computing I learned from sigfpe" T-shirt * OlegFacts: Oleg can evaluate bottom. With his fists. * quicksilver: my computer starts to play 'Dies Irae' when shapr gets ops, automatically. About the Haskell Weekly News New editions are posted to [20]the Haskell mailing list as well as to [21]the Haskell Sequence and [22]Planet Haskell. [23]RSS is also available, and headlines appear on [24]haskell.org. To help create new editions of this newsletter, please see the information on [25]how to contribute. Send stories to byorgey at cis dot upenn dot edu. The darcs repository is available at darcs get [26]http://code.haskell.org/~byorgey/code/hwn/ . References 1. http://haskell.org/ 2. http://fantasticcontraption.com/ 3. http://article.gmane.org/gmane.comp.lang.haskell.cafe/45786 4. http://article.gmane.org/gmane.comp.lang.haskell.cafe/45586 5. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/OneTuple 6. http://article.gmane.org/gmane.comp.lang.haskell.libraries/10134 7. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/protocol-buffers 8. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/protocol-buffers… 9. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hprotoc 10. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/45634 11. http://thread.gmane.org/gmane.comp.lang.haskell.glasgow.user/15339 12. http://thread.gmane.org/gmane.comp.lang.haskell.libraries/10133 13. http://planet.haskell.org/ 14. http://haskell.org/haskellwiki/Blog_articles 15. http://luqui.org/blog/archives/2008/10/03/laziness-and-the-monad-laws/ 16. http://mult.ifario.us/p/the-haskell-platform-and-lessons-learned-elsewhere 17. http://koweycode.blogspot.com/2008/10/darcs-weekly-news-6.html 18. http://lindstroem.wordpress.com/2008/09/27/inheritance-in-composites-and-ov… 19. http://billsix.blogspot.com/2008/10/dabbling-with-haskell.html 20. http://www.haskell.org/mailman/listinfo/haskell 21. http://sequence.complete.org/ 22. http://planet.haskell.org/ 23. http://sequence.complete.org/node/feed 24. http://haskell.org/ 25. http://haskell.org/haskellwiki/HWN 26. http://code.haskell.org/~byorgey/code/hwn/

1 0

An important question to Simon P. Jones
by Vag 04 Oct '08

04 Oct '08

Dear Simon! Can you please give an authoritative explanation, in the nutshell, what is exact relationship between object-oriented programming and proposed lightweight extensible records? Thanks in advance, sincerely yours, Vag.

1 0

NewLines in string
by Serge D. Mechveliani 04 Oct '08

04 Oct '08

People, Here is a certain naive question about string setting in a Haskell program. I have the functions parseProgram :: String -> Program apply :: Program -> Term -> Term -- interpreter, and try to build the example modules for them as in the following example: ---------------------------------------------------------------- module Main where program = parseProgram " isOrdered nil -> true; isOrdered (X:nil) -> true; isOrdered (X:Y:Ys) -> isOrd (X > Y) (isOrdered (Y:Ys)); isOrd true bo -> false; isOrd false bo -> bo " main = let {data = ... :: Term; res = apply program data} in putStr $ concat ["data = ", shows data "\n", "res = ", shows res "\n"] ---------------------------------------------------------------- -- that is set a string (an object program example), parse it to a program, apply this program and print the result. But in reality, the above string must be set like this: (2) "\ \ isOrdered nil -> true;\ \ isOrdered (X:nil) -> true;\ \ isOrdered (X:Y:Ys) -> \ \ isOrd (X > Y) (isOrdered (Y:Ys));\ \\ \ isOrd true bo -> false;\ \ isOrd false bo -> bo\ \" -- because otherwise Haskell treats the NewLine characters in the above string in an unneeded way. But with (2), the program Main occurs hard to read. One could try to improve this by using readFile. Form the file data.txt -------------------------------- isOrdered nil -> true; isOrdered (X:nil) -> true; isOrdered (X:Y:Ys) -> isOrd (X > Y) (isOrdered (Y:Ys)); isOrd true bo -> false; isOrd false bo -> bo ----------------------------------------- and combine readFile "data.txt", the `do' construct, pearseProgram, apply, putStr. But this needs adding one more file, and this also changes the programming style in defining examples. Now, what is the difference? 1. readFile reads string from file, with treating NewLine as needed. 2. The Haskell compiler reads the expression parseProgram " isOrdered nil > true; ... " from the input Haskell program with treating NewLine in the string in a different way. I wonder: how to remove most of backslashes in the program (2) ? Thank you in advance for your comments. ----------------- Serge Mechveliani mechvel(a)botik.ru

2 1