- Haskell-Cafe - Haskell.org

Haskell with ncurses
by Atrudyjane 21 Nov '16

21 Nov '16

As a beginning Haskeller working my way through the Haskell book, I'm looking ahead at different options for a user interface. The idea is to build various small tools for personal use. After reading around, it looks like web apps are preferred. Tried to use the Atom editor, and it just seemed a little slow on my machine. The interfaces don't need to be complex or "pretty", just clean looking. I would almost prefer something simple like ncurses. What are your opinions/experiences with using ncurses with Haskell? Or is that the wrong path altogether? Regards, Andrea Sent with [ProtonMail](https://protonmail.com) Secure Email.

4 4

[cabal-install bootstrap failed] ghc-pkg: missing --global-package-db option, location of global package database unknown
by CRUMEYROLLE Pierre 21 Nov '16

21 Nov '16

hello i try to reinstall cabal in a customized directory but bootstrap launch fail because ghc-pkg list is failed thanks for help $ ghc --version The Glorious Glasgow Haskell Compilation System, version 7.10.3 $ ./bootstrap.sh --no-doc Using gcc for C compiler. If this is not what you want, set CC. Using /usr/libexec/gcc/x86_64-redhat-linux/4.8.5/collect2 instead. Checking installed packages for ghc-7.10.3... ghc-pkg: missing --global-package-db option, location of global package database unknown Error during cabal-install bootstrap: running 'ghc-pkg list' failed $ ghc-pkg list ghc-pkg: missing --global-package-db option, location of global package database unknown

1 0

Investigating Evaluation, Typing and Deduction -- WIP
by Rik Howard 21 Nov '16

21 Nov '16

Comments welcome. http://www.dcs.bbk.ac.uk/~rik/gallery/investigating-evaluation-typing-and-d…

1 1

Am I misunderstanding something about `StableName` and `hashStableName`?
by raichoo 20 Nov '16

20 Nov '16

Hi, I've been playing around with `System.Mem.StableName` and `hashStableName` and stumbled across a behavior that seems to be rather weird. First of all the documentation of `hashStableName` says the following: The Int returned is not necessarily unique; several StableNames may map to the same Int (in practice however, the chances of this are small, so the result of hashStableName makes a good hash key). OK, sounds fine, let's put this to the test. So I wrote a little program. module Main where import Control.Monad import System.Mem.StableName import System.Environment main :: IO () main = do args <- getArgs res <- forM [0..10000] $ \i -> do x <- makeStableName i let h' = hashStableName x unless (null args) $ print h' return h' putStrLn "---------------------------" print $ minimum res print $ maximum res Running this program without an argument only prints aus the minimum and maximum hash values that where produced. So surprises here: $ ./Test --------------------------- 1 10000 Here is how things get a bit weird when I make the progam print out the hash values while they are being produced I get a completely different result. Not only that but I get a whole lot of hash collisions, since the hashes produced only seem to be within the range of 1 and 260. $ ./Test print [...] 128 127 126 125 124 123 122 121 120 119 118 117 116 115 114 --------------------------- 1 260 Am I missing something fundamental here, because this behavior seems to be rather confusing. Kind regards, raichoo

3 2

Esqueleto maintainership
by Christopher Allen 20 Nov '16

20 Nov '16

Cross-posting my intent to take over maintainership of Esqueleto. Can read more here: https://github.com/haskell-infra/hackage-trustees/issues/84 This has gone on for awhile, but it's beginning to eat into maintenance of my work projects and it seems like there are others eager to contribute and help share the load where there wasn't 6 or 12 months ago. Cheers, Chris Allen

3 3

difficulty writing GEq instance
by Anders Papitto 20 Nov '16

20 Nov '16

Hello! I'm having a lot of trouble writing a Data.GADT.Compare.GEq instance - can anyone help me fill in the blank? For context - I'm generating Tag types automatically for using with DSum, and I need a GEq instance. That's part of an attempt to add efficient rendering of sum types to Reflex/Reflex-Dom (https://github.com/anderspapitto/reflex-sumtype- render/blob/master/src/ReflexHelpers.hs). I put this question on stackoverflow as well a day ago ( http://stackoverflow.com/questions/40698207/how-can-i- write-this-geq-instance). Here's the code (it's a full, standalone file - you can copy it into Foo.hs and run ghc to see the full error I'm facing). The error I get is that when I try to recursively call geq on the unwrapped x and y, I can't because ghc considers them to have different types - Quux a and Quux b. However, the whole point of why I'm trying to call geq is to see if a and b are the same, so I'm pretty confused. Note that I'm making use of the generics-sop library, which is where NP and NS and I come from. {-# LANGUAGE GADTs #-} {-# LANGUAGE DataKinds #-} {-# LANGUAGE TypeOperators #-} {-# LANGUAGE KindSignatures #-} {-# LANGUAGE RankNTypes #-} module Foo where import Data.GADT.Compare import Generics.SOP import qualified GHC.Generics as GHC data Quux i xs where Quux :: Quux (NP I xs) xs newtype GTag t i = GTag { unTag :: NS (Quux i) (Code t) } instance GEq (GTag t) where -- I don't know how to do this geq (GTag (S x)) (GTag (S y)) = let _ = x `geq` y in undefined

2 2

How to profile haskell web applications?
by Moritz Angermann 19 Nov '16

19 Nov '16

Hi, I’m looking to (perf, heap) profile a haskell web application. With ekg and the ekg wai middleware I can get some data on the rts, gc, and latency. What’s the approach towards performance (-p) and heap (-h) profiling for web applications? Cheers, moritz

1 0

GHC plugin: custom wired-in package, module, and type class
by J. Stutterheim 18 Nov '16

18 Nov '16

Hi all, I'm looking for a way to add my own wired-in package/module/type class for use in a core-to-core plugin. My first challenge is to identify the type class that I want to find. I have the following code: myPkgUnitId :: UnitId myPkgUnitId = fsToUnitId (fsLit "mypackage") mkMyModule :: FastString -> Module mkMyModule m = mkModule myPkgUnitId (mkModuleNameFS m) myModule :: Module myModule = mkMyModule (fsLit "My.Module") myClassName :: Name myClassName = clsQual myModule (fsLit "MyClass") myClassKey myClassKey :: Unique -- How do I get this? Here the missing piece is the myClassKey. Looking at how this is done for, e.g., the Monad class: monadClassKey :: Unique monadClassKey = mkPreludeClassUnique 8 we see that it has a hard-coded unique identifier. Is there a way to do the same for my own type class? Or to somehow find the Unique key dynamically? The second challenge is more or less the same in that I want to identify a member of MyClass: myFunMName :: Name myFunMName = varQual myModule (fsLit "myFun") myFunMClassOpKey myFunMClassOpKey :: Unique -- How do I get this? Again, the identifier is hard-coded for, e.g., ">>=". Any insights will be much appreciated. Best regards, Jurriën

2 1

4.4.2 Fixity Declarations - and what happens when they are missing
by Johannes Waldmann 18 Nov '16

18 Nov '16

Yeah, I know, it's bikeshedding, some 30 years after the fact, but - How on Earth was this allowed to happen? "Any operator lacking a fixity declaration is assumed to be infixl 9" https://www.haskell.org/onlinereport/haskell2010/haskellch4.html#x10-820061 If I don't specify associativity or precedence for an operator, then the parser should simply reject any expression whose meaning would depend on such declarations. So I'd have to write parentheses at each (ambiguous) use site - or write a fixity declaration. Both would serve to document the code. The current situation feels like "the type of identifiers without type declaration is int". Short of repairing the standard - I'd welcome GHC to print a huge warning for each definition and use (!) of "operator without fixity decl". - J. PS: to show that I did at least some research * there's no mention of this in 4.2 "Infix operators" in http://research.microsoft.com/~simonpj/papers/history-of-haskell/history.pdf * Scala also has default rules (6.12.3) "The precedence of an infix operator is determined by the operator’s first character." "Operators ending in a colon ‘ : ’ are right-associative. All other operators are left-associative." and why do I care at all? This: https://github.com/jwaldmann/haskell-obdd/issues/1

1 0

Preventing dereferencing an STRef after taking an action
by Timotej Tomandl 18 Nov '16

18 Nov '16

Hi, is it possible to prevent an STRef passed to one ST action from being passed further on to another ST action? I am trying to prevent these usages: invalid :: STRef s a -> STRef s b -> ST s () invalid x y=action1 x y >>=action2 x y invalid1 :: STRef s a -> STRef s b -> ST s () invalid1 x y=action1 x y >>=action2 y while still being able to do this: valid :: STRef s a -> STRef s b -> ST s () valid x y=action1 x y >>=action2 x where action1 will modify x and y inplace, but in the end only x will hold the useful information. In other words I am trying to make it impossible for any action after action1 to have a same argument as action1's second argument. timo

1 0