January 2014 - Haskell-Cafe

Re: [Haskell-cafe] Licenses and dependencies
by Mike Meyer 12 Jan '14

12 Jan '14

On Sat, Jan 11, 2014 at 2:17 PM, Mateusz Kowalczyk <fuuzetsu(a)fuuzetsu.co.uk> wrote: > Hey, > Perhaps you should have CC'd the list as well for this reply. Oops. I've moved it back, since you suggested it. > On 11/01/14 20:04, Mike Meyer wrote: > >> On 11/01/14 18:44, Ben Foppa wrote: > >> Once you distribute the binary, you have probably pulled in and used the > >> licensed libraries (unless replaced) so you're subject to the strongest > >> license used. > > > > Not "strongest", but all. Well, the GPL licenses sort of "stack", > > since they explicitly allow you to use newer (and usually stronger) > > versions of themselves, but other license don't. > > > > The GNU project provides a description of many of the available > > licenses at: http://www.gnu.org/licenses/license-list.html, including > > compatibility notes. > Right, all, but what this usually entices is that the strongest one > prevails. For example, code with BSD3 and GPLv3 sources will have to > be released under GPLv3 as a whole (although each part is still what > it was licensed under individually). I worked under the assumption > that all dependencies are have compatible licenses (and their > dependencies too &c). No, it's not a contest. You *have* to abide by them all. If you're dealing with the GPL, it may look like the "strongest" prevails, but that's a property of the GPL. It specifically requires any derived work be distributed under the GPL, with no added restrictions. So a compatible license will have no extra restrictions, and appear "weaker". A "stronger" license with extra restrictions (like the 4-clause BSD license) will be incompatible with the GPL, so you can't distributed a derived work at all. If you take the GPL out of the mix, and wanted to distributed a work derived from something covered by (for instance) BSD4 and the CPL - well, that would be fine, because I don't think either prevents the restrictions in the other. But the derived work would be covered by both, not just the stronger of the two (whichever that is). Your usual case may well be GPL'ed code. I happen to work with BSD-licensed code bases more often than not. > >> I think you'll be fine just choosing the PublicDomain license option. [...] > >> Things you can't do with such an option: > >> > >> * bundle other people's code which isn't under public domain > > > > This is vague. If your bundle is a single work (like a static binary), > > then it has to obey all the applicable licenses, and probably can't be > > in the public domain. If it's a collection of works, then it can > > include anything you want, so long as each element in the collection > > obeys it's license. > > I meant bundle as in the repository/source tarball/however you publish > your source, _not_ the binary (I explicitly mention binary somewhere else). In that case, you're perfectly free to include sources with other licenses in the tarball. You just need to make sure you follow their license for their entry in the collection. It's well-established that a simple collection of works (a tarball, etc.) is not a derived work. > >> * somehow enforce that only certain libraries will run with your > >> program: the user should be able to replace the BSD3 licensed > >> dependencies with their own if they want to > > Is there any license that enforces that? That seems more like a > > technical issue than a legal one. Though if you're dynamically linking > > the libraries, things get incredibly confusing. > LGPL, does it not? It even outlines what constitutes as a prevention > of other libraries. I believe the LGPL does just the opposite - it *permits* linking with non-free code without creating a derived work. I think I may not be understanding what you're saying, though. I think you're implying that it's possible to force users to use specific libraries. That is indeed the case - you could write a modified version of the GPL that says "You may only redistribute works derived from my project if they use my libraries". You are correct in that you can't do that if you place the work in the public domain. On the other hand, I don't know of any commonly used open source library that does that. <mike

4 5

different behaviours with or without putStrLn
by Corentin Dupont 12 Jan '14

12 Jan '14

Hi guys, I'm experimenting different behaviours with or without a "putStrLn"! :( Basically, with the following code, I want the evaluation to really happen on the "evaluate". I found out that it doesn't: it is evaluated elsewhere (I don't know where). If I put a putStrLn (commented below), the evaluation really happens there. *execCommand :: (TVar MyState) -> StateT MyState IO () -> IO ()execCommand ts sm = do s <- atomically $ readTVar ts s' <- execStateT sm s s'' <- evaluate s' --evaluation should happen here, but it doesn't --putStrLn $ displayMulti $ _multi s'' atomically $ writeTVar ts s''* To give you more context, I have a state that, when evaluated, might not terminate. So I added a watchdog (like in mueval), that will kill the thread in case the evaluation doesn't terminate. That's why I need to be sure of where the evaluation takes place. Thanks! Corentin

5 6

Parse Error ( Parsec )
by mukesh tiwari 12 Jan '14

12 Jan '14

Hello Cafe, I am trying to write a parser for propositional logic[1]. It's working fine for every input except equivalence ( <=> ). *Main> calculator "a=>b" Imp (Lit 'a') (Lit 'b') *Main> calculator "a<=b" Red (Lit 'a') (Lit 'b') *Main> calculator "a<=>b" *** Exception: failed to parse I think, the reason is parser taking equivalence ( <=> ) as reduction ( <= ) and next character is '>' so it is parse error . If I remove both implication and reduction then equivalence is working fine. *Main> calculator "a<=>b" Eqi (Lit 'a') (Lit 'b') Could some please tell me how to solve this problem. I also tried fixity declaration but got this error LogicPraser.hs:12:10: The fixity signature for `<=>' lacks an accompanying binding -Mukesh Tiwari [1] http://logic.stanford.edu/classes/cs157/2010/notes/chap02.html {-# LANGUAGE NoMonomorphismRestriction #-} import Text.Parsec.Token import Text.Parsec.Prim import Text.Parsec.Char import Text.Parsec.Expr import Text.Parsec.Combinator import Text.Parsec.Language import Control.Applicative hiding ( ( <|> ) , many ) import Data.Maybe ( fromJust ) --infixl 9 <=> data LExpr = Lit Char | Not LExpr | And LExpr LExpr | Or LExpr LExpr | Imp LExpr LExpr -- (=>) | Red LExpr LExpr -- ( <= ) | Eqi LExpr LExpr -- ( <=> ) deriving Show exprCal = buildExpressionParser table atom table = [ [ Prefix ( Not <$ string "~" ) ] , [ Infix ( And <$ string "&" ) AssocLeft ] , [ Infix ( Or <$ string "|" ) AssocLeft ] , [ Infix ( Imp <$ string "=>" ) AssocLeft , Infix ( Red <$ string "<=" ) AssocLeft , Infix ( Eqi <$ string "<=>" ) AssocLeft ] ] atom = char '(' *> exprCal <* char ')' <|> ( Lit <$> letter ) calculator :: String -> LExpr calculator expr = case parse exprCal "" expr of Left msg -> error "failed to parse" Right ( val ) -> val ~

4 4

TypeLits and ScopedTypeVariables
by Ben Foppa 11 Jan '14

11 Jan '14

I've been playing with TypeLits recently, trying to create a vector with type-fixed size. I've hit a stumbling block trying to compile this (simplified) code: {-# LANGUAGE DataKinds #-} {-# LANGUAGE KindSignatures #-} import Control.Exception (assert) import GHC.TypeLits data Foo (n :: Nat) a = Foo a size :: SingI n => Foo n a -> Int size = fromInteger . fromSing . sizeSing where sizeSing :: Foo n a -> Sing n sizeSing _ = sing The error is: Could not deduce (SingI Nat n1) arising from a use of `sing' from the context (SingI Nat n) bound by the type signature for size :: SingI Nat n => Foo n a -> Int at /home/ben/test.hs:10:9-33 Possible fix: add an instance declaration for (SingI Nat n1) In the expression: sing In an equation for `sizeSing': sizeSing _ = sing In an equation for `size': size = fromInteger . fromSing . sizeSing where sizeSing :: Foo n a -> Sing n sizeSing _ = sing Can anybody shed light on this? It seems like the kind of thing that ScopedTypeVariables should solve, but it doesn't make a difference.

2 2

Licenses and dependencies
by Ben Foppa 11 Jan '14

11 Jan '14

Many permissive licenses (I'm specifically looking at MIT and BSD3) include a clause that the license file must be included in distributions of the source or the binary. It seems to me that if I have a project on Github, that qualifies as redistribution of the source. My questions are thus: Do I need to include the license files from all my dependencies in my Github repository? Does having them listed as dependencies in my .cabal file "implicitly" include them? Does making a project "PublicDomain" only refer to the code contained in that project, or would it (unintentionally and unlawfully) give distributors of that source code the freedom to not include the BSD3/MIT license files from dependencies? Essentially my goal is to waive all intellectual property rights to most of my Haskell projects, to the extent that, were I to unintentionally sign away my intellectual property, my open-source contributions would be safe - what's the easiest way to do this? Thanks, Ben Foppa

3 3

ANN: xxHash, fast, high quality, non-cryptographic checksums.
by christian＠ponies.io 10 Jan '14

10 Jan '14

I've just finished implementing a haskell version of the xxHash algorithm, which is documented briefly here (mostly in code): https://code.google.com/p/xxhash/ The haskell version is *almost* as fast as C and allows the generation of a better quality checksum than adler32, in less time. Benchmarks are here: http://ponies.io/posts/2013-01-10-xxhash.html Code is here: https://github.com/christian-marie/xxhash Critique welcome!

1 0

Slow mvar when compiled with threaded
by Branimir Maksimovic 09 Jan '14

09 Jan '14

I have test network client, something like apache bench tool. It uses mvars to synchronize and everything is ok when compiled without -threaded. real 0m2.995s user 0m0.601s sys 0m2.391s With -threaded compile option result is following: real 0m18.196s user 0m2.054s sys 0m3.313s Seems that program is sleeping most of the time for some reason. I can't explain behavior as it seems that program is ok. It starts `concurrency` threads which wait on mvar to process next task. Program follows: {-# Language OverloadedStrings #-} import System.CPUTime import System.IO --import System.IO.Error import Network.Socket hiding(recv) import Network.Socket.ByteString import System.Environment import Control.Concurrent import Control.Exception main = do n <- getArgs let (host,port,conc,reqs) = parse n putStrLn $ "Connecting to " ++ host ++ " " ++ port s <- getAddrInfo Nothing (Just host) (Just port) let servAddr = head s begin <- getCPUTime process servAddr conc reqs end <- getCPUTime let diff = (fromIntegral (end - begin))/(10^12) :: Double putStrLn $ show (round (fromIntegral reqs / diff)) ++ " r/s" parse [h,p,conc,reqs] = (h,p,read conc::Int,read reqs::Int) parse _ = error "usage client host port concurrency requests" process servAddr conc reqs = do let niter = if reqs >= conc then conc else reqs putStrLn $ "loop " ++ show niter mvars <- initThreads niter [] putStrLn $ "Initialized " ++ show niter let loop n (m:mvs) f | n>0 = do flag <- isEmptyMVar m if f > length mvars then putStrLn "busy" else return () if flag || f > length mvars then do putMVar m () loop (n-1) mvs 0 else loop n mvs (f+1) | otherwise = return () loop n [] f = if n>0 then loop n mvars f else return () putStrLn $ "length " ++ show (length mvars) loop (reqs-niter) mvars 0 where initThreads niter vars | niter > 0 = do mvar <- newMVar () forkIO $ process mvar initThreads (niter-1) (mvar:vars) | otherwise = return vars process mvar = do sock <- socket (addrFamily servAddr) Stream defaultProtocol connect sock (addrAddress servAddr) sendAll sock "Hello World!\n" buf <- recv sock 1024 close sock takeMVar mvar process mvar

2 8

Re: [Haskell-cafe] Why Maybe exists if there is Either?
by Niklas Larsson 09 Jan '14

09 Jan '14

Maybe doesn't have to be used for error reporting. Optional values is a very big use case for it. ----- Ursprungligt meddelande ----- Från: "Vlatko Basic" <vlatko.basic(a)gmail.com> Skickat: ‎2014-‎01-‎09 16:37 Till: "haskell-cafe(a)haskell.org" <haskell-cafe(a)haskell.org> Ämne: Re: [Haskell-cafe] Why Maybe exists if there is Either? Hi Daniel, > If you just want to signalize a fail case without any additional > information, then a Maybe fits better than an Either, because > why should you need this bogus empty string? Now it looks to me that it might be better and more consistent to write an empty bogus string (as with: nothing = Left "") than to have two distinct ways of error reporting, and both are used widely and often should be intermixed. And at the end, call site could decide does it want to use the string or not. vlatko -------- Original Message -------- Subject: Re: [Haskell-cafe] Why Maybe exists if there is Either? From: Daniel Trstenjak <daniel.trstenjak(a)gmail.com> To: haskell-cafe(a)haskell.org Date: 09.01.2014 16:13 > > Hi Vlatko, > > On Thu, Jan 09, 2014 at 03:50:16PM +0100, Vlatko Basic wrote: >> I'm curious to find out what was the reasoning to make Maybe? >> What is the added value with introducing it? >> In which situations the above substitution does not hold? > > If you just want to signalize a fail case without any additional > information, then a Maybe fits better than an Either, because > why should you need this bogus empty string? > > How should you know that the string doesn't contain something relevant? > > > Greetings, > Daniel > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe > _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

4 3

How to determine the right path to haddock html documentation?
by Carlo Hamalainen 09 Jan '14

09 Jan '14

Hi, I'm writing a patch for ghcmod-vim so that a user can look up the haddock html documentation for a symbol. My code makes three system calls, e.g. to look up "Just" it does: $ ghc-mod info foo.hs Foo Just data Maybe a = ... | Just a -- Defined in `Data.Maybe' $ ghc-pkg find-module Data.Maybe --simple-output haskell2010-1.1.1.0 base-4.6.0.1 $ ghc-pkg field base-4.6.0.1 haddock-html haddock-html: /home/carlo/opt/ghc-7.6.3_build/share/doc/ghc/html/libraries/base-4.6.0.1 Then "Data.Maybe" is changed to Data-Maybe.html" and appended to the haddock-html value. But this procedure doesn't work in general, for example when looking up String: $ ghc-mod info foo.hs Foo String type String = [Char] -- Defined in `GHC.Base' $ ghc-pkg find-module GHC.Base /home/carlo/opt/ghc-7.6.3_build/lib/ghc-7.6.3/package.conf.d base-4.6.0.1 /home/carlo/.ghc/x86_64-linux-7.6.3/package.conf.d $ ghc-pkg field base-4.6.0.1 haddock-html haddock-html: /home/carlo/opt/ghc-7.6.3_build/share/doc/ghc/html/libraries/base-4.6.0.1 $ ls /home/carlo/opt/ghc-7.6.3_build/share/doc/ghc/html/libraries/base-4.6.0.1/GHC-Base.html ls: cannot access /home/carlo/opt/ghc-7.6.3_build/share/doc/ghc/html/libraries/base-4.6.0.1/GHC-Base.html: No such file or directory because the html is in the src subdirectory: $ ls /home/carlo/opt/ghc-7.6.3_build/share/doc/ghc/html/libraries/base-4.6.0.1/src/GHC-Base.html /home/carlo/opt/ghc-7.6.3_build/share/doc/ghc/html/libraries/base-4.6.0.1/src/GHC-Base.html But this isn't even the right place to go, one should look at the Prelude file http://hackage.haskell.org/package/base-4.6.0.1/docs/Prelude.html#t:String which is what Hoogle suggests. How does Hoogle get it right? It looks like haskell-mode for emacs runs into a similar problem: https://github.com/haskell/haskell-mode/blob/master/inf-haskell.el#L734-L740 ------------------------------- (defvar inferior-haskell-ghc-internal-ident-alist ;; FIXME: Fill this table, ideally semi-automatically. '(("GHC.Base.return" . "Control.Monad.return") ("GHC.List" . "Data.List"))) (defun inferior-haskell-map-internal-ghc-ident (ident) "Try to translate some internal GHC identifier to its alter ego in haskell docs." ------------------------------- Thanks, -- Carlo Hamalainen http://carlo-hamalainen.net

3 9

Altering Class hierarchy
by Frantisek Farka 09 Jan '14

09 Jan '14

Hi all, I was toying with Haskell classes and came across Functor-Applicative-Monad proposal[1]. It shows out that adding a super class (Applicative) to a class (Monad) is not an easy task when the class hierarchy is already in use. In this case it result in stepwise approach (step1 e. g. [2]) in order not to break already existing code which uses Monad. I am interested in other examples of such changes. Which alterings of your class hierarchy do you usually need when rewriting your code? I don't mean only adding superclass here, other modifications could be also handy, e. g. dropping out a class or splitting a class into two. Can you point me to some actual examples on real source code? How do you deal with this kind of problem on your projects? Thanks for a reply :) FF [1] http://www.haskell.org/haskellwiki/Functor-Applicative-Monad_Proposal [2] https://ghc.haskell.org/trac/ghc/ticket/8004

2 1