March 2012 - Haskell-Cafe

Summer of Code idea: Haskell Web Toolkit
by Dimitry Golubovsky 07 Mar '12

07 Mar '12

Hi, Alejandro Serrano Mena wrote: > My idea is to make a client-side Haskell Web Toolkit, in the spirit of > Google Web Toolkit, which would allow to program in Haskell the client part > of a web application, and would complement the web frameworks already > existing for Haskell (such as Yesod and Snap). http://www.haskell.org/haskellwiki/Haskell_web_toolkit - In memoriam ;) Feel free to use these ideas. It would be nice if you could pick it up and complete - I never did... -- Dimitry Golubovsky Anywhere on the Web

1 0

Re: [Haskell-cafe] [Haskell] Higher types in contexts
by Henning Thielemann 06 Mar '12

06 Mar '12

moved to haskell-cafe On Mon, 5 Mar 2012, Barney Hilken wrote: > Is there any deep reason why I can't write a polymorphic type in a context? I think the record update problem can be (sort of) solved if you could write: > > class Has r Rev (forall a. [a] -> [a]) => HRClass r where > setHRClass :: (forall a.[a] -> [a]) -> r -> r > > but polymorphic types are not allowed in contexts. Is this one of the problems SPJ considers "Hard" or is it a feasible extension? I don't know what you want to do, but you may wrap the (forall a. [a] -> [a]) in an existential type: data ListFunc = forall a. ListFunc ([a] -> [a]) class Has r Rev ListFunc => HRClass r where setHRClass :: ListFunc -> r -> r

4 3

Re: [Haskell-cafe] Adding type annotations to an AST?
by oleg＠okmij.org 06 Mar '12

06 Mar '12

> How do I add type annotations to interior locations in an abstract > syntax tree?.... > i.e. the type it [algorithm] infers is the type of the whole > program, I would also like the types of any internal let-rec > definitions so I can label my AST. I had exactly the same problem: type reconstruction and the annotation of all sub-terms with their inferred types. Even if the overall type inference fails, the user can still see what the type checker was able to infer before the error. Here is the solution http://okmij.org/ftp/Computation/FLOLAC/TEvalNR.hs There is a bit of the explanation here: http://okmij.org/ftp/Computation/FLOLAC/lecture.pdf

2 1

Double-dispatch
by Clark Gaebel 06 Mar '12

06 Mar '12

Is there any way in Haskell to have the correct function selected based on the types of two different types? For example, let's say I'm writing intersection tests: aABBandAABB :: AABB -> AABB -> Bool oBBandOBB :: OBB -> OBB -> Bool oBBandPoint :: OBB -> Point -> Bool Is there some way (such as with Type Families) that I can write some sort of generic method for this: intersects :: Intersectable -> Intersectable -> Bool which automatically selects the right function to call based on the types of the two intersectables? Regards, - clark

3 4

Conduit Sink fork
by t helfferich 05 Mar '12

05 Mar '12

Hi! So, it turns out I have a need for a sink that forks input into two other sinks using the Conduit package. Here is what I came up with: https://gist.github.com/1975383 Is this the right way to write it? Also, what left over value(s) should I return in the Done constructor? I appreciate any help you can give me. Thanks so much,Grant

2 6

Reasons for Super-Linear Speedup
by burak ekici 05 Mar '12

05 Mar '12

Dear List; I have parallelized RSA decryption and encryption schemes by using second generation strategies of Haskell p.l. In that case, what I got in the sense of speed up was nearly 10 times of better performances (on a quad-core CPU with 8M cache) in the parallel evaluation of 125K long plain text with 180-bit of an encryption key, in comparison with the serial evaluation, abnormally. All my thoughts are in the direction of getting wrong serial-time performances. My question here is that is there any difference between serial and parallel evaluation of any arbitrary computation in terms of "cache usage methodologies" in Haskell compiler design. If no; what could be other possible reasons? I would be much more than appreciated, if somebody brings an acceptable pinpoint for the issue. Many thanks in advance. Kind regards Burak.

2 1

[haskell-cafe] Question about 64bit target on Windows platform
by C K Kashyap 05 Mar '12

05 Mar '12

Hi All, Can someone please let me know if there is a 64bit target on Windows on the horizon for GHC? I am trying to push for changing the current implementation in my organization in C++ to Haskell - Our primary targets are Windows and Mac. Not being able to generate 64bit DLL's on Windows would be a big bottleneck. Regards, Kashyap

3 4

Looking for an heap-like data structure allowing efficient update
by Arnaud Bailly 05 Mar '12

05 Mar '12

Hello Cafe, I am looking for a data structure that would have the following informal properties: - allow efficient retrieval of minimum element for some ordering relation on a computed property of the elements - allow efficient update of any element wrt to this property I have first tried to use a PQueue from fingertree package but it does not allow deletion/update of arbitrary elements. I then started to try combining this with a map but ran into the same issue. I skimmed through Okasaki's book (oldies but goldies...) and the basic finger tree structure but these does not seem to fit my needs. Thanks for your help Regards, Arnaud Bailly

2 2

Re: [Haskell-cafe] FreeSect -- generalised sections syntax extension
by Ras Far 05 Mar '12

05 Mar '12

So we can have ... chiastic freesects? This sounds like too much fun! I find chiasmus is a term from linguistics? From [1]: "The elements of simple chiasmus are often labelled in the form ABBA, where the letters correspond to grammar, words, or meaning." So we've got love, too (Aramaic word “Abba” = love of God or so...) Also coincidental, since a lot of the literary instances are apparently found in sacred texts (although Coleridge comes up a lot too, cough). So many coincidences around this. I'd written up the idea for the Haskell extension in 2003 (calling them "arbitrary sections"), but only decided to implement them about a week ago. Had just gotten started about an hour earlier, when eyebloom on #haskell asks about the same thing, right down to the _ syntax. (I offered him collaboration but he didn't get back to me, hope you're not sore eyebloom, anyhow I don't mind you were maybe too busy.) Then I get hacking Language.Haskell.Exts and -- the most recent extension already added is "TupleSections"! And there's no other extension so related to this idea. Now the serendipity with Wren's linguistic research. And even the freegeek domain. Fantastic. Wren, your slides look great, thanks for the link, I think I'll read them tonight. Do you have a paper version of the slides at all? Slides can be a bit terse on their own. Yours Chiastically, Andrew [1] http://en.wikipedia.org/wiki/Chiastic_structure

3 3

Impact of "try" on Parsec performance
by Omari Norman 04 Mar '12

04 Mar '12

The Parsec documentation says that Parsec performs best on predictive grammars, and Parsec does not backtrack by default to improve performance (though this also improves error messages). On the other hand, I notice that attoparsec and polyparse backtrack by default, and attoparsec claims to be faster than Parsec (I can't remember if polyparse makes this claim). The Parsec documentation says there is a penalty for using try and suggests left-factoring grammars when possible. Real World Haskell says that excessive use of try can degrade Parsec performance quite significantly. On the other hand, some of the combinators provided with Parsec, such as notFollowedBy, use try. So my question is: what is the practical impact of using try? I ask because as a novice I have a simple grammar that could, perhaps, be factored to avoid use of try, but it's quite a brain puzzle for me and I wonder if it would ever be worth it. Does it matter how many characters "try" would have to store, or how often it would run?

7 7