May 2014 - Haskell-Cafe

ANN: fsnotify-0.1
by Roman Cheplyaka 15 May '14

15 May '14

A new version of fsnotify, a cross-platform filesystem notification library, is available. http://hackage.haskell.org/package/fsnotify-0.1.0.1 The main user-visible changes since the last version are: * ability to stop a listening job * ability to set a polling interval (when polling is used) * a couple of small semantic changes (see [changelog] for details) * improved documentation Plus, there was a fair bit of internal refactoring, including a new test suite. [changelog]: https://github.com/haskell-fswatch/hfsnotify/blob/master/CHANGELOG.md There's still a long-standing bug about wrong event types being generated on OS X. We know how to fix it (basically, use kqueue instead of OS X's own API), but I don't have time to do it. So if anyone is interested, help would be welcome! See https://github.com/haskell-fswatch/hfsnotify/issues/36 Thanks to Miguel Mitrofanov for providing access to an OS X host, which was extremely valuable while working on and testing this version. Roman

1 0

case, GADTs, Arrows
by Miguel Mitrofanov 15 May '14

15 May '14

Hi Café! Are there EXACT rules for desugaring "case" construct in Arrow syntax? The documentation says just "The translation is similar to that of if commands". But that can't be right in case of GADTs. For example, let's say we have data A where A :: Show s => s -> A Then the following would work fine: proc () -> do a <- arr A -< "test" case a of A s -> returnA -< show s But if we abstract part of it: b :: (Arrow a, Show s) => a s String b = proc(s) -> returnA -< show s proc() -> do a <- arr A -< "test" case a of A s -> b -< s then it fails. Now, I'm not asking why exactly does it fail. I understand, that type classes are just data types, and, if arrow is fixed, then the real type of "b" would be b :: Show s -> a s String and that can't be combined with something of type a () (Show s, s). So, the second example shouldn't work at all (although the error message could use some improvement). But I'm surprised that the first one does. Another unexpected thing is that it still requires ArrowChoice from the arrow, although there are no binary alternatives here. Therefore, the rules would be quite helpful.

3 2

Reading and writing to handles made with System.Process far too slow
by Mateusz Kowalczyk 13 May '14

13 May '14

Hi, I'm have some business in piping some data and reading some data back out of a socket so I thought that I'd just use the ‘socat’ tool. I went off to System.Process just to find out that reading and writing are taking far too long. I put together a small example which only requires that you have ‘cat’ on your system: {-# LANGUAGE UnicodeSyntax #-} module Uzbl.WithSource where import GHC.IO.Handle ( hPutStr, hGetContents, hSetBuffering , BufferMode(..)) import System.Process ( createProcess, proc , StdStream(CreatePipe), std_out, std_in) gs ∷ IO String gs = do let sp = (proc "cat" []) { std_out = CreatePipe, std_in = CreatePipe } (Just hin, Just hout, _, _) ← createProcess sp -- hSetBuffering hin NoBuffering -- hSetBuffering hout NoBuffering hPutStr hin "Test data" hGetContents hout All this should effectively do is to give you back "Test data". While it *does* do that, it takes far too long. When I run ‘gs’, it will start to (lazily) print the result, printing nothing but opening ‘"’ and then after about 2-3 seconds printing the rest and finishing. If we set buffering on the in-handle (hin) to NoBuffering, we get a slightly different behaviour: pretty much straight away we'll have ‘"Test data’ but then it will wait for the same amount of time to conclude that it's the end of the response. Changing buffering mode on ‘hout’ seems to make no difference. Setting precise number in a BlockBuffering seems to be no improvement and in the actual application I will not know how long the data I'm piping in and out will be. GHC 7.8.2, process-1.2.2.0; I'm running ‘gs’ in GHCi. It seems that if I change the module name to Main, make ‘main = gs >>= putStrLn’, compile the file and run it, it just hangs there! If I add a newline at the end, it will print but the program will not finish. This makes me think that perhaps I should be closing handles somewhere (but if I try inside the function, I get no output, thanks lazy I/O). What I would expect this program to do is to produce same result as ‘print "Test data" | cat’. -- Mateusz K.

3 3

ANNOUNCE: graphviz-2999.17.0.0
by Ivan Lazar Miljenovic 13 May '14

13 May '14

It's taken me over a year, but I'm pleased to announce the latest version of my graph visualisation (via the Graphviz suite of tools) library: http://hackage.haskell.org/package/graphviz You can read all the changes in the Changelog (now linked to from the Hackage page, thanks to the magic of Hackage 2.0! :o), but here are some explicit non-API changes that people should probably be made aware of: * Parsing is currently about 8x slower than previously... because I suggested something two years ago to Malcolm Wallace on what seemed to make polyparse faster; he then released that change as part of polyparse-1.9 a year ago, but now when I do proper benchmarks (rather than the extremely limited ones I did then) I find that my changes actually make it _slower_ in real world usage. Whoops! I'm still using polyparse-1.9 in the hopes that Malcolm is able to release a point-fix release to revert my suggestion. * I'm not sure when it happened, but upstream Graphviz seems to have changed the behaviour of `dot -Tcanon` such that nodes are now interspersed among the edges (and I have been told that it's just a pretty-printer and that it's behaviour should not matter). As such, it is highly recommended that people only use the Generalised representation for parsing unless they're very sure of their input sources. One other change that people might find useful: if I ever again take so long (i.e. 4 major version releases of upstream) to update graphviz such that the various Attribute definitions are no longer valid and fail to parse, there's now (semi-experimental) support for having unrecognised attributes be parsed as an UnknownAttribute instead; but most of the other functions aren't aware of this functionality and thus you might need to duplicate a lot of code if you want to use this with other features like round-tripping. -- Ivan Lazar Miljenovic Ivan.Miljenovic(a)gmail.com http://IvanMiljenovic.wordpress.com

1 0

FLOPS 2014 call for participation (early registration deadline extended)
by Eijiro Sumii 13 May '14

13 May '14

Dear colleagues, The early registration for FLOPS 2014 (Twelfth International Symposium on Functional and Logic Programming, June 4-6, Kanazawa, Japan) is now open until May 16 Friday, 23:59 JST. - Please register *now* (that is, if you plan to attend) at: http://www.jaist.ac.jp/flops2014/registration.html - Program: http://www.jaist.ac.jp/flops2014/program.html - Invited talks by Ranjit Jhala, Shin-ya Katsumata, and Gabriele Keller: http://www.jaist.ac.jp/flops2014/invited.html - Proceedings to be published as LNCS 8475 - Hyakuman-goku Matsuri Festival http://www.kanazawa-tourism.com/eng/event/event2.php on the day after the symposium (June 7); accommodation can be extended on request basis via the registration site (use the comments field) - More information at: http://www.jaist.ac.jp/flops2014/ Thank you again, FLOPS 2014 Program Co-Chairs Michael Codish and Eijiro Sumii

1 0

Visiting Portland 6/12-6/14 (and Eugene after that)
by Joachim Breitner 13 May '14

13 May '14

Hi, I’ll be visiting a summer school in Eugene¹ in June and my travel plans include two days in Portland, (Tuesday June 12th, afternoon until Saturday, June 14th, shortly past noon). If there anybody wants to meet for keysigning/sightseeing/discussing Haskell/other stuff, let me know. Also, I’d be happy to join any interesting event such as User group meetings, and I can offer to hold a talk (most likely something Haskell related, or general Debian topics). I haven’t book accommodation yet. Useful secret tips are welcome :-) After that I’ll be in Eugene for two weeks and although the schedule is full I’m sure there will be time to meet likewise people in the evening. Or even during the event – I would be surprised if I were there the only haskell-cafe reader that attends the summer school. Greetings, Joachim ¹ http://www.cs.uoregon.edu/research/summerschool/summer14/ -- Joachim “nomeata” Breitner mail(a)joachim-breitner.de • http://www.joachim-breitner.de/ Jabber: nomeata(a)joachim-breitner.de • GPG-Key: 0xF0FBF51F Debian Developer: nomeata(a)debian.org

1 0

Error when trying to use Hashable
by John Ky 12 May '14

12 May '14

Hi Haskell Cafe, I have some code that compiles when running on GHC on OS X, but not on Ubuntu: No instance for (hashable-1.2.1.0:Data.Hashable.Class.GHashable (GHC.Generics.Rep Point)) arising from a use of `hashable-1.2.1.0:Data.Hashable.Class.$gdmhashWithSalt' Possible fix: add an instance declaration for (hashable-1.2.1.0:Data.Hashable.Class.GHashable (GHC.Generics.Rep Point)) In the expression: (hashable-1.2.1.0:Data.Hashable.Class.$gdmhashWithSalt) In an equation for `hashWithSalt': hashWithSalt = (hashable-1.2.1.0:Data.Hashable.Class.$gdmhashWithSalt) In the instance declaration for `Hashable Point' Anyone know what's happening here? Cheers, -John -- <http://www.gocatch.com/> Sydney, Australia <https://www.facebook.com/goCatch> <https://twitter.com/gocatchapp> <http://www.linkedin.com/company/goCatch> <https://itunes.apple.com/au/app/gocatch/id444439909?mt=8> <https://play.google.com/store/apps/details?id=com.gocatchapp.goCatch&hl=en> <http://www.windowsphone.com/en-au/store/app/gocatch/d76b0eb5-bad6-429f-b99e…> <http://appworld.blackberry.com/webstore/content/31917887/>

2 2

what did I do wrong
by Roelof Wobben 12 May '14

12 May '14

6 6

cabal install lens fails
by John Ky 11 May '14

11 May '14

Hi Haskell Cafe, Anyone know about this error? Installing library in /Users/johnky/Library/Haskell/ghc-7.6.3/lib/reflection-1.4/lib Registering reflection-1.4... Installed reflection-1.4 Configuring mtl-2.0.1.0... Building mtl-2.0.1.0... Preprocessing library mtl-2.0.1.0... [ 1 of 21] Compiling Control.Monad.Writer.Class ( Control/Monad/Writer/Class.hs, dist/build/Control/Monad/Writer/Class.o ) [ 2 of 21] Compiling Control.Monad.State.Class ( Control/Monad/State/Class.hs, dist/build/Control/Monad/State/Class.o ) [ 3 of 21] Compiling Control.Monad.Reader.Class ( Control/Monad/Reader/Class.hs, dist/build/Control/Monad/Reader/Class.o ) [ 4 of 21] Compiling Control.Monad.RWS.Class ( Control/Monad/RWS/Class.hs, dist/build/Control/Monad/RWS/Class.o ) [ 5 of 21] Compiling Control.Monad.Identity ( Control/Monad/Identity.hs, dist/build/Control/Monad/Identity.o ) [ 6 of 21] Compiling Control.Monad.Error.Class ( Control/Monad/Error/Class.hs, dist/build/Control/Monad/Error/Class.o ) Control/Monad/Error/Class.hs:93:18: Not in scope: `catch' Failed to install mtl-2.0.1.0 Cheers, -John -- <http://www.gocatch.com/> Sydney, Australia <https://www.facebook.com/goCatch> <https://twitter.com/gocatchapp> <http://www.linkedin.com/company/goCatch> <https://itunes.apple.com/au/app/gocatch/id444439909?mt=8> <https://play.google.com/store/apps/details?id=com.gocatchapp.goCatch&hl=en> <http://www.windowsphone.com/en-au/store/app/gocatch/d76b0eb5-bad6-429f-b99e…> <http://appworld.blackberry.com/webstore/content/31917887/>

2 2

Custome monoid mappend on HashMap and HashSet values
by John Ky 11 May '14

11 May '14

Hello, Does anyone have an example on how I might be able to use monoids to mappend a HashMap Int (HashMap Int Int) together? Cheers, -John -- <http://www.gocatch.com/> Sydney, Australia <https://www.facebook.com/goCatch> <https://twitter.com/gocatchapp> <http://www.linkedin.com/company/goCatch> <https://itunes.apple.com/au/app/gocatch/id444439909?mt=8> <https://play.google.com/store/apps/details?id=com.gocatchapp.goCatch&hl=en> <http://www.windowsphone.com/en-au/store/app/gocatch/d76b0eb5-bad6-429f-b99e…> <http://appworld.blackberry.com/webstore/content/31917887/>

4 5