- Haskell-Cafe - Haskell.org

funny comics about Monad
by sajeeb Bhm 22 Feb '16

22 Feb '16

check this one!

2 1

Re: [Haskell-cafe] Manual type-checking in graphs: Avoidable?
by Gesh 21 Feb '16

21 Feb '16

On February 19, 2016 10:54:11 PM GMT+02:00, Francesco Ariis <fa-ml(a)ariis.it> wrote: >hey Gesh, > >you are right (not able to compile it atm too, but it looks >correct and way elegant). >Maybe post it in the Ml to help OP? > >ciao ciao >F > > >On Fri, Feb 19, 2016 at 04:59:56PM +0200, Gesh wrote: >> I'm away from compiler at the moment, but... >> Shouldn't this work? >> > {-# LANGUAGE GADTs #-} >> > data NodeS = HamsterS | PersonS >> > data NodeP a where >> > Hamster :: String -> NodeP HamsterS >> > Person :: String -> NodeP PersonS >> > data Node = forall a. NodeP a >> > type Graph = Gr Node... >> > hamsters :: NodeP PersonS -> ... >> >> Basically the idea of that you reify the choice of constructor to the >type level, permitting static restriction of the constructors used. >> >> HTH, >> Gesh Oops, meant to send to list.

6 9

ANN: stack-1.0.4
by Emanuel Borsboom 20 Feb '16

20 Feb '16

See haskellstack.org for installation and upgrade instructions. Major changes: - Some notable changes in stack init: - Overall it should now be able to initialize almost all existing cabal packages out of the box as long as the package itself is consistently defined. - Choose the best possible snapshot and add extra dependencies on top of a snapshot resolver rather than a compiler resolver - #1583 <https://github.com/commercialhaskell/stack/pull/1583> - Automatically omit a package (--omit-packages) when it is compiler incompatible or when there are packages with conflicting dependency requirements - #1674 <https://github.com/commercialhaskell/stack/pull/1674>. - Some more changes for a better user experience. Please refer to the doc guide for details. - Add support for hpack, alternative package description format #1679 <https://github.com/commercialhaskell/stack/issues/1679> Other enhancements: - Docker: pass ~/.ssh and SSH auth socket into container, so that git repos work #1358 <https://github.com/commercialhaskell/stack/issues/1358> . - Docker: strip suffix from docker —version. #1653 <https://github.com/commercialhaskell/stack/issues/1653> - Docker: pass USER and PWD environment bariables into container. - On each run, stack will test the stack root directory (~/.stack), and the project and package work directories (.stack-work) for whether they are owned by the current user and abort if they are not. This precaution can be disabled with the --allow-different-user flag or allow-different-user option in the global config (~/.stack/config.yaml). #471 <https://github.com/commercialhaskell/stack/issues/471> - Added stack clean --full option for full working dir cleanup. - YAML config: support Zip archives. - Redownload build plan if parsing fails #1702 <https://github.com/commercialhaskell/stack/issues/1702>. - Give mustache templates access to a ‘year’ tag #1716 <https://github.com/commercialhaskell/stack/pull/1716>. - Have “stack ghci” warn about module name aliasing. - Add “stack ghci —load-local-deps”. - Build Setup.hs with -rtsopts #1687 <https://github.com/commercialhaskell/stack/issues/1687>. - stack init accepts a list of directories. - Add flag infos to DependencyPlanFailures (for better error output in case of flags) #713 <https://github.com/commercialhaskell/stack/issues/713> - stack new --bare complains for overwrites, and add --force option #1597 <https://github.com/commercialhaskell/stack/issues/1597>. Bug fixes: - Previously, stack ghci would fail with cannot satisfy -package-id when the implicit build step changes the package key of some dependency. - Fix: Building with ghcjs: “ghc-pkg: Prelude.chr: bad argument: 2980338” #1665 <https://github.com/commercialhaskell/stack/issues/1665>. - Fix running test / bench with --profile / --trace. - Fix: build progress counter is no longer visible #1685 <https://github.com/commercialhaskell/stack/issues/1685>. - Use “-RTS” w/ profiling to allow extra args #1772 <https://github.com/commercialhaskell/stack/issues/1772>. - Fix withUnpackedTarball7z to find name of srcDir after unpacking (fixes stack setup fails for ghcjs project on windows) #1774 <https://github.com/commercialhaskell/stack/issues/1774>. - Add space before auto-generated bench opts (makes profiling options work uniformly for applications and benchmark suites) #1771 <https://github.com/commercialhaskell/stack/issues/1771>. - Don’t try to find plugin if it resembles flag. - Setup.hs changes cause package dirtiness #1711 <https://github.com/commercialhaskell/stack/issues/1711>. - Send “stack templates” output to stdout #1792 <https://github.com/commercialhaskell/stack/issues/1792>. ------------------------------ Thanks to all our contributors for this release: - Aaron Wolf @wolftune <https://github.com/wolftune> - Artem Chernyak @achernyak <https://github.com/achernyak> - Dan Aloni @da-x <https://github.com/da-x> - Daniel Gasienica @gasi <https://github.com/gasi> - Emanuel Borsboom @borsboom <https://github.com/borsboom> - Erik Stevenson @narrative <https://github.com/narrative> - Gauthier Segay @smoothdeveloper <https://github.com/smoothdeveloper> - Greg Weber @gregwebs <https://github.com/gregwebs> - Grégoire Détrez @gdetrez <https://github.com/gdetrez> - Harendra Kumar @harendra-kumar <https://github.com/harendra-kumar> - Mark Karpov @mrkkrp <https://github.com/mrkkrp> - Martin Kolinek - Matt Gambogi @gambogi <https://github.com/gambogi> - Michael Sloan @mgsloan <https://github.com/mgsloan> - Michael Snoyman @snoyberg <https://github.com/snoyberg> - Oleg Grenrus @phadej <https://github.com/phadej> - Eldar Yakupov @panamiga <https://github.com/panamiga> - Prayag Verma @pra85 <https://github.com/pra85> - Simon Hengel @sol <https://github.com/sol> - Simon Jakobi @sjakobi <https://github.com/sjakobi> - Tristan Webb @drwebb <https://github.com/drwebb> - Yves Parès @YPares <https://github.com/YPares> And thanks also to the 150+ contributors to past releases!

1 0

Re: [Haskell-cafe] Manual type-checking in graphs: Avoidable?
by Johannes Waldmann 20 Feb '16

20 Feb '16

> .. a graph that described which people own which hamsters, This is a bipartite graph, or, a relation. You want a type like data Rel src tgt a = Rel { fore :: ! ( M.Map src (M.Map tgt a) ) , back :: ! ( M.Map tgt (M.Map src a) ) } where src = People, tgt = Hamster, and a = Bool. In general, a could be some extra weight information. - J.W. cf. https://gitlab.imn.htwk-leipzig.de/waldmann/pure-matchbox/blob/master/src/M…

5 6

Manual type-checking in graphs: Avoidable?
by Jeffrey Brown 19 Feb '16

19 Feb '16

I use FGL, which (roughly) defines type Gr a b as a graph on nodes of type a and edges of type b. Suppose you wanted a graph that described which people own which hamsters, knowing only their name. You would have to make node and edge types like this: data GraphNode = Person String | Hamster String data GraphEdge = Has where the strings represent their names. Suppose then you wanted to write a function that, given a person, returns the names of all their hamsters. To make sure the call makes sense, the function would have to first check that the input is in fact a person. Since persons and hamsters are both constructors of the same type, you can't let Haskell's robust, beautiful type-checking system distinguish them for you; you've got to write something like "case n of Person _ -> True; _ -> False". Is there some way around writing such manual checks? -- Jeffrey Benjamin Brown

5 6

Munich Haskell Meeting, 2015-02-23 @ 19:30
by Christian Neukirchen 19 Feb '16

19 Feb '16

Dear all, Next week, our monthly Munich Haskell Meeting will take place again on Tuesday, February 23 at Cafe Puck at 19h30. For details see here: http://chneukirchen.github.io/haskell-munich.de/dates If you plan to join, please add yourself to this dudle so we can reserve enough seats! It is OK to add yourself to the dudle anonymously or pseudonymously. https://dudle.inf.tu-dresden.de/haskell-munich-feb-2016/ Everybody is welcome! cu, -- Christian Neukirchen <chneukirchen(a)gmail.com> http://chneukirchen.org

1 0

Confusing behavior in Haskell networking libraries
by Alex Rozenshteyn 19 Feb '16

19 Feb '16

I was trying to write a web scraper, so I used scalpel <https://hackage.haskell.org/package/scalpel>. The website I wanted to scrape blocks my IP (I run a tor exit node), so I decided to use proxychains <http://proxychains.sourceforge.net/> (specifically, version 3.1-6 according to Debian). I ran into the following weird behavior: if I tell proxychains to run dns through the proxy, things are fine, but if I tell it to run dns in the clear or the URL I'm trying to connect to is an IP address (e.g. manually resolved), I always get timeouts (much faster than I should). (don't resolve dns over the proxy) % proxychains stack exec -- test-scalpel "http://ifconfig.co" ProxyChains-3.1 (http://proxychains.sf.net) |R-chain|-<>-201.175.94.245:38746-<><>-188.113.88.193:80-<--timeout (resolve dns over the proxy) % proxychains stack exec -- test-scalpel "http://ifconfig.co" ProxyChains-3.1 (http://proxychains.sf.net) |DNS-request| ifconfig.co |R-chain|-<>-201.175.111.245:10000-<><>-4.2.2.2:53-<><>-OK |DNS-response| ifconfig.co is 188.113.88.193 |R-chain|-<>-201.175.111.245:10000-<><>-188.113.88.193:80-<><>-OK (resolve dns over proxy, but use an IP to avoid actually doing it) % proxychains stack exec -- test-scalpel "http://188.113.88.193" ProxyChains-3.1 (http://proxychains.sf.net) |R-chain|-<>-201.175.111.245:10000-<><>-188.113.88.193:80-<--timeout curl does not have this behavior: % proxychains stack exec -- curl "http://188.113.88.193" ProxyChains-3.1 (http://proxychains.sf.net) |R-chain|-<>-201.172.16.131:55599-<><>-188.113.88.193:80-<><>-OK % proxychains stack exec -- curl "http://ifconfig.co" ProxyChains-3.1 (http://proxychains.sf.net) |DNS-request| ifconfig.co |R-chain|-<>-201.172.17.231:10000-<><>-4.2.2.2:53-<><>-OK |DNS-response| ifconfig.co is 188.113.88.193 |R-chain|-<>-201.172.17.231:10000-<><>-188.113.88.193:80-<><>-OK % vim proxychains.conf # to change the setting % proxychains stack exec -- curl "http://ifconfig.co" ProxyChains-3.1 (http://proxychains.sf.net) |R-chain|-<>-201.172.16.131:55599-<><>-188.113.88.193:80-<><>-OK wget and aria2c also behave like curl. wreq behaves like scalpel: % proxychains stack exec -- test-wreq "http://188.113.88.193" ProxyChains-3.1 (http://proxychains.sf.net) |R-chain|-<>-201.172.17.231:10000-<><>-188.113.88.193:80-<--timeout HTTP (the Haskell package) behaves differently than all the rest, failing to connect even where the rest succeed: % proxychains stack exec -- test-http "http://ifconfig.co" ProxyChains-3.1 (http://proxychains.sf.net) |DNS-request| ifconfig.co |R-chain|-<>-201.175.94.245:38746-<><>-4.2.2.2:53-<><>-OK |DNS-response| ifconfig.co is 188.113.88.193 |R-chain|-<>-201.175.94.245:38746-<><>-188.113.88.193:80-<--timeout Gists of the programs I ran: https://gist.github.com/rpglover64/f668ed372c63e271cf15 Anyone have any idea what's going on?

1 0

Closed classes
by David Feuer 19 Feb '16

19 Feb '16

I was thinking about ~two notions of "closed class" yesterday, and I'm curious if anyone's done any work on either concept. In each case, the class definition is followed by *all* of its instances, and the instances are checked *in order* (rather than based on specificity and OVERLAPS/OVERLAPPABLE pragmas). No backtracking: If the instance head matches, GHC commits to it. Associated types are treated as closed type families, and would work just the same (I don't think any significant extension to the closed type family mechanism would be required). This seems to make a very nice parallel to the usual open classes with open associated types. And it lets you combine overlapping instances with associated types without (I believe) risking type safety. Backtracking: GHC does not commit to the instance until it has satisfied the instance constraints. This lets instance writers offer multiple alternative instance constraints. Associated types would be a good bit trickier. One option would be to require all instances with the same head to share a type/data instance. The other (much more invasive) option would be to allow the instance chosen to guide the type selection, which would push the backtracking into the type checker.

5 10

Call for Haskell.org committee self-nominations
by John Wiegley 19 Feb '16

19 Feb '16

Dear Haskellers, It is time to put out the call for new volunteers to the haskell.org committee. We have one member due for retirement since this past October: Gershom Bazerman. The committee would like to thank him for his excellent service. To nominate yourself, please send an email to committee at haskell.org by March 4, 2016. The retiring members are eligible to re-nominate themselves. Please feel free to include any information about yourself that you think will help us to make a decision. Being a member of the committee does not necessarily require a significant amount of time, but committee members should aim to be responsive during discussions when the committee is called upon to make a decision. Strong leadership, communication, and judgement are very important characteristics for committee members. The role is about setting policy, providing direction/guidance for Haskell.org infrastructure, planning for the long term, and being fiscally responsible with the Haskell.org funds (and donations). As overseers for policy regarding the open source side of Haskell, committee members must also be able to set aside personal or business related bias and make decisions with the good of the open source Haskell community in mind. More details about the committee's roles and responsibilities are on https://wiki.haskell.org/Haskell.org_committee If you have any questions about the process, please feel free to e-mail us at committee at haskell.org, or contact one of us individually. Thank you, -- John Wiegley GPG fingerprint = 4710 CF98 AF9B 327B B80F http://newartisans.com 60E1 46C4 BD1A 7AC1 4BA2

1 0

STM unperformance
by Wojtek Narczyński 19 Feb '16

19 Feb '16

Dear list, I create four haskell threads, each performs disjoint STM transactions. The more system threads I add, the slower the program runs. With four system threads the performance is 14% of the nonparallel run. What is going on? $ ghc -v Glasgow Haskell Compiler, Version 7.10.3, stage 2 booted by GHC version 7.8.4 $ ghc -O2 -threaded test-stm.hs $ time ./test-stm +RTS -N1 real 1.843s $ time ./test-stm +RTS -N2 real 7.469s $ time ./test-stm +RTS -N3 real 9.640s $ time ./test-stm +RTS -N4 real 13.144s $ cat test-stm.hs {-# LANGUAGE ScopedTypeVariables #-} import Control.Monad import Control.Concurrent import Control.Concurrent.STM import Control.Concurrent.STM.Stats import System.IO.Unsafe -- Copied from GHC docs children :: MVar [MVar ()] children = unsafePerformIO (newMVar []) waitForChildren :: IO () waitForChildren = do cs <- takeMVar children case cs of [] -> return () m:ms -> do putMVar children ms takeMVar m waitForChildren forkChild :: IO () -> IO ThreadId forkChild io = do mvar <- newEmptyMVar childs <- takeMVar children putMVar children (mvar:childs) forkFinally io (\_ -> putMVar mvar ()) -- Test case main = do forkChild $ newTVarIO 0 >>= incrManyTimes "thread1" forkChild $ newTVarIO 0 >>= incrManyTimes "thread2" forkChild $ newTVarIO 0 >>= incrManyTimes "thread3" forkChild $ newTVarIO 0 >>= incrManyTimes "thread4" waitForChildren dumpSTMStats -- Confirms no conflicts incrManyTimes :: String -> TVar Int -> IO () incrManyTimes l = incrRec (1000000 :: Int) where incrRec n v | n == 0 = pure () | otherwise = trackNamedSTM l (modifyTVar v (+1)) >> incrRec (n-1) v -- Wojtek

2 2