May 2012 - Haskell-Cafe

Haskell Hackathon in Munich
by Heinrich Hördegen 02 May '12

02 May '12

Dear all, the Munich Haskell Meeting is glad to announce a local one day Hackathon in Munich. It will take place the 12th of May 2012 from 10am to 6pm. Please checkout the details at: http://www.haskell-munich.de/dates Cordially yours, Heinrich -- -- hoerdegen(a)funktional.info www.funktional.info --

2 1

Reader monad, refactoring and missing the point all at once
by Eugene Dzhurinsky 02 May '12

02 May '12

Hi all! Last day I was trying to fix idiii library, because it uses utf8 for parsing non-unicode content. I found the functions > -- | Parses one value and returns it as a 'String' > parseString :: CharEncoding -> TagParser String > parseString enc = do > v <- case enc of > 0x01 -> parseUntilWord16Null -- UTF-16 > 0x02 -> parseUntilWord16Null -- UTF-16 BOM > _ -> parseUntilWord8Null -- ISO-8859-1 or UTF-8 > return $ encPack enc v > > encPack :: CharEncoding -> [Token] -> String > encPack 0x00 s = Text.unpack $ decodeASCII $ BS.pack s > encPack 0x01 (0xFF:0xFE:s) = Text.unpack $ decodeUtf16LE $ BS.pack s > encPack 0x01 (0xFE:0xFF:s) = Text.unpack $ decodeUtf16BE $ BS.pack s > encPack 0x02 s = Text.unpack $ decodeUtf16BE $ BS.pack s > encPack _ s = Text.unpack $ decodeUtf8 $ BS.pack s updated the dependency from > import Data.Text.Encoding (decodeASCII, decodeUtf16LE, decodeUtf16BE, decodeUtf8) to > import Data.Text.ICU.Convert and added implementation for decoding functions: > decodeAny :: String -> BS.ByteString -> Text.Text > decodeAny charset src = unsafePerformIO $ ((flip toUnicode) src) `fmap` open charset (Just True) > > decodeASCII :: BS.ByteString -> Text.Text > decodeASCII = decodeAny "latin1" > > decodeUtf16LE = decodeAny "utf-16le" > > decodeUtf16BE = decodeAny "utf-16be" > > decodeUtf8 = decodeAny "utf-8" Now I want to add possibility to specify encoding to yse with decodeASCII. I was thinking of adding Reader monad and providing some sort of charset configuration there - but it will lead up to complicating the code, which uses this parseString function. And this code is used inside Parser of Text.ParserCombinators.Poly.State - so I will need to update all usages of this parser. Another approach might be to use IORef with encoding stored there, but I don't really like this solution. What would be the best way of refactoring of such kind? Thanks! -- Eugene N Dzhurinsky

1 0

Creating a dependency graph (in the hope it may show some structure)
by Bram Neijt 02 May '12

02 May '12

Dear Haskelllers, I've reacently started with Haskell and during the Dutch Hackaton[3], I decided to generate graphs of dependencies. I was thinking about doing this for Java earlier, and that weekend I decided to write the tooling in Haskell. I've written two programs: hs2dot[1] and dotimate[2]. The first generates a dot file (Graphviz) from all .hs files below the current working directory. The second will take one or more dot files and call neato (from Graphviz) multiple times to generate frames of neato output from one to the next. Trying hs2dot and dotimate on two sequencial commits in the cabal/cabal-install/frames directory already gives me a way to complex and full graph. See attached. ==> My question is: What dependencies, if any, would be nice to graph, and/or which dependencies could I filter out to keep this from becoming a mess? Greetings, Bram Neijt [1] https://gist.github.com/bneijt/hs2dot [2] https://gist.github.com/bneijt/dotimate [3] http://www.haskell.org/haskellwiki/DHD_UHac#Utrecht_Hackathon

2 2

ANN: bitwise - fast multi-dimensional unboxed bit packed Bool arrays
by Claude Heiland-Allen 01 May '12

01 May '12

Hi all, I'm pleased to announce bitwise-0.1: http://hackage.haskell.org/package/bitwise --8<-- excerpt from the hackage page Unboxed multidimensional bit packed Bool arrays with fast aggregate operations based on lifting Bool operations to bitwise operations. There are many other bit packed structures out there, but none met all of these requirements: 1. unboxed bit packed Bool array, 2. multi-dimensional indexing, 3. fast (de)serialization, or interoperable with foreign code, 4. fast aggregate operations (fold, map, zip). --8<-- end excerpt bitwise has been tested in hugs -98, ghc-7.0.4, ghc-7.4.1. Bool data structures comparison (features numbered as excerpt above): package features additional notes 1 2 3 4 array Y Y n n bitarray Y n n ? based on array bitstream n n ? ? based on vector bitstring Y n ? ? based on bytestring bit-vector n n Y ? based on vector bitwise Y Y Y Y repa n Y Y ? based on vector uvector Y n ? ? deprecated vector n n Y ? uses a Word8 for each Bool Known shortcomings of this first release of bitwise: * missing common instances (Eq, Ord, Read, Show, ...); * missing additional functions (counting, searching, ...); * missing documentation on ByteString format: the bits are taken from the array and packed into Word8s, with the first bit (in array order) becoming the least significant bit of the first Word8; the last Word8 is padded with 0 at the most-significant-bit end; * PBM reading is not compliant to the official specification; * misleading benchmark (bitwise zipWith is fast, but there may be a faster way to zipWith UArray ix Bool than the one I used). What I'm using bitwise for at the moment: A building block for fractal imaging: an image is made up of cells, each cell can be in one of 4 states (represented with 2 bits): * cell is 100% interior to a set of points in the plane * cell is 100% exterior * partially interior and partially exterior (ie, on the boundary) * unknown with recursive subdivision of cells into images taking place only when the cell is on the boundary of the set being imaged. Thanks, Claude

1 0

Re: [Haskell-cafe] Darcs home page updated
by Paul R 01 May '12

01 May '12

On Mon, 30 Apr 2012 12:27:57 -0700, Simon Michael <simon(a)joyful.com> said: Simon> With 2.8 released, I felt Darcs deserves better presentation. After Simon> surveying other VCS sites I worked on an update to our home page Simon> layout and content over the last few days, with review and input from Simon> #darcs, and it went live last night. It's far from perfect but I hope Simon> it's a good step forward. Thanks for the input, and have at it: Simon> http://darcs.net Hi, that looks good, thanks. Here are some ideas : - The purple color of titles is not very attractive, changing it to something more saturated would make it look fresher. - the purple color rounded box is not very attractive either :) - there is a video right on the first page, good ! Unfortunatly, it is not showing darcs but its cousin. Also, the title is "why do we continue to develop ... [camp]", which isn't the most positive approach one can expect on a start page - the compact tutorial is great, I love it. It deserves more poish. For inspiration, one can have a look at this website for example : http://gembundler.com/ - underlined links looks a bit 90's. You can add a simple css rule to choose an other color and remove the underline (though it is good practice to set it on mouseover) cheers, -- Paul

2 2

ANN: HandsomeSoup-0.3: CSS selectors for HXT
by aditya bhargava 01 May '12

01 May '12

*Homepage:* http://egonschiele.github.com/HandsomeSoup *On Hackage:* http://hackage.haskell.org/package/HandsomeSoup *Blurb:* HandsomeSoup is the library I wish I had when I started parsing HTML in Haskell. It is built on top of HXT and adds a few functions that make is easier to work with HTML. Most importantly, it adds CSS selectors to HXT. The goal of HandsomeSoup is to be a complete CSS2 parser for HXT (it is very close to this right now). -- adit.io

2 2