September 2014 - Haskell-Cafe

Re: [Haskell-cafe] usage of traversal
by Michael Snoyman 23 Sep '14

23 Sep '14

On Tue, Sep 23, 2014 at 10:12 AM, Kazu Yamamoto <kazu(a)iij.ad.jp> wrote: > Hi Michael, > > Probably, you forgot to Cc: to haskell-cafe. > > > I'd be -1 on an operator, I think having a named function for this is a > > good thing for readability of code. > > > > As far as good style: I personally think it is. In classy-prelude, I > > actually export the Foldable-based `mapM` by default, and will regularly > > use that (or forM) for this kind of code. > > With AMP, 'fmap' and 'liftM' are identical and we use <$> instead > recently. Likewise, 'traverse' and 'mapM' are identical. If we > introduce an operator, say <:>, we can forget 'traverse' and 'mapM' > when writing code and can write: > > getModificationTime <:> mfile > > This looks like function application like Applicative style. > > Just a thought. > > --Kazu > It's definitely true that with AMP the situation will be much improved, and it makes more sense to use an operator here since it will cover what is today two distinct functions. However, I think I'm overall still in favor of sticking to the named function, since both `mapM` and `traverse` are well known. This may just be a chicken-and-egg argument, however, and once <:> takes hold it's just as readable as `mapM`. Michael

2 2

ANNOUNCE: bv-0.3.0
by Iago Abal 23 Sep '14

23 Sep '14

I am pleased to announce a new release of the bv package, a library for bit-vector arithmetic. http://hackage.haskell.org/package/bv This is essentially a maintenance release to support newest versions of GHC and base. Yet, I took the opportunity to do some code and documentation clean up. A changelog is available at https://bitbucket.org/iago/bv-haskell/src/v0.3.0/CHANGES.md. Iago

1 0

usage of traversal
by Kazu Yamamoto 23 Sep '14

23 Sep '14

Hi, Recently, I need to write the following code: getModTime :: Maybe FilePath -> IO (Maybe UTCTime) getModTime mfile = case mfile of Nothing -> return Nothing Just file -> Just <$> getModificationTime file I feel that this is redundant. So, I used 'traverse' instead: getModTime :: Maybe FilePath -> IO (Maybe UTCTime) getModTime mfile = getModificationTime `traverse` mfile First, I would like to know whether or not this is a good coding style. Second, if this is acceptable, why don't we define an operator? For instance, (<:>) :: (Traversable t, Applicative f) => (a -> f b) -> t a -> f (t b) (<:>) = traverse getModTime :: Maybe FilePath -> IO (Maybe UTCTime) getModTime mfile = getModificationTime <:> mfile Is there such an operator already? Regards, --Kazu

2 1

Bound and free variables
by Jan Stolarek 22 Sep '14

22 Sep '14

Hi *, I have a simple question about terminology regarding bound and free variables. Assume I have: let f x = let g y = ... in g c in ... Now: - `c` is free in `g` and `f` - `y` is bound in `g` - `x` is free in `g`. - `x` is bound in `f` What about `y` in `f`? Is it also bound in `f`? If so then it certainly is bound in a different way that `x`. Is there a terminology that allows to distinguish these different forms of bound variables? Janek

5 6

[ANN] codex 0.2.0.0
by Alois Cochard 22 Sep '14

22 Sep '14

Hi Cafe, I released a new version of my tool `codex`[1] (a ctags file generator for cabal project dependencies). This new version include: - By default, the current project is now included in the generated `codex.tags` file - Resolve dependencies in the "benchmarks" of your cabal project - Fix tags file header to respect specification (to make them work in Sublime) - Cache eviction mechanism (no need of using `--force` in default use cases) - Add default configuration for Sublime Thanks to everyone for bug fixes and feature requests! Cheers [1] https://github.com/aloiscochard/codex -- *Λ\ois* http://twitter.com/aloiscochard http://github.com/aloiscochard

1 0

how to use hs_init_ghc and pass runtime params to a dll
by Kees Bleijenberg 22 Sep '14

22 Sep '14

Can someone provide a (pointer to a) working basic example of C code that uses a Windows dll created by ghc that calls hs_init_ghc? Can someone confirm that hs_init_ghc uses stdcall in a windows dll. Is ghc -no-hs-main -shared -threaded -O2 -o TBGlas.dll TBGlasDll.hs a correct way to create the dll? Do I have to use the ghc parameter -dynamic? if I use this parameter, the ghc tells me to install the 'dyn' libraries for package base. Sounds aweful. I'am using ghc version 7.8.3. and the 32 bits Haskell Platform on 64 bits Windows 7. Background: I create a dll with ghc. The program that uses the dll starts with a lookup for all functions I want to use from the dll. The result is a bunch of function pointers. I use a function pointer to call the real function. Then I call the defaultRtsConfig function from the dll. I get a pointer to a struct with a byte for the enumarable RTSOptEnabled followed by a pointer to a string. In the returned struct this pointer is nil. Then I set the pointer in the struct to the string "-N -s". After that I call hs_init_ghc(nil,nil,defaultRtsConcfigStruct). This works fine. I get a overview of the sparks created, the elapsed times, there is a speedup from the parallel code etc. But sometimes (1 out of 100) the program crashes. Any ideas what could be wrong? Kees

1 0

ghc-prof-options and libraries on Hackage
by Oliver Charles 22 Sep '14

22 Sep '14

Hi all, I'd like to start a conversation around the ghc-prof-options option that can be used in Cabal-ized libraries. As you may know, this option specifies extra GHC options that are used when --enable-library-profiling is enabled. This is very convenient for local development, but I argue that it can be counter-productive for releases onto Hackage. To provide an example, I'm currently working on a little game engine that uses JuicyPixels to load images. I have a problem in my code that needs optimizing, but the current state of things results in profiles that are very difficult to work with. JuicyPixels specifies -auto-all in its cabal file, which means I have no alternative but to profile JuicyPixels code. In this scenario, the bottleneck is actually within my FRP game loop and nothing to do with image loading! As a result, the profiles are fairly useless to me. Roman Cheplyaka also points out that by doing this, profiles are skewed - -auto-all is not free, so we actually pay in runtime performance every time a library does this. I would like to hear from others if we should consider managing this option a little more. My personal feeling is that this flag shouldn't be used unless in local development, so guarded by an off-by-default build flag. I think cabal should also warn authors who are using this flag, and encourage them to place a guard around this option. Thoughts? - ocharles

9 8

Haskell job opportunity in Santiago, Chile
by Anupam Jain 22 Sep '14

22 Sep '14

Hi all, We are looking to hire an experienced Haskell developer for our startup in Santiago, Chile. This will be a 6-8 week project involving web development using Yesod and you will be building a key component of our product. Part time is OK. We would prefer someone who can work out of our office, however, working remote is also a possibility. If remote, being in India would be a plus. Good english communication ability is essential. Please email me directly if interested. Thanks, Anupam Jain

1 0

DSL implementation
by felipe zapata 22 Sep '14

22 Sep '14

HI all, as a part of a project for physical simulation, I'm was given the task to help with the implementation of an Small DSL that is a minimal functional language. Point is that this small language contains some dependent typed theory, because we want to be able to check statically the shapes of vector and matrices, among other things. Haskell of course seems the tool for the job. But my boss insists that the implementation should be done in C, but I think that it is madness. Can someone please point me to relevant literature, discussions or general information that can help me to show my boss that Haskell is the right tool? any help would be appreciated. Felipe Z.

6 6

Fwd: Pointers by using functions
by Holden Lee 21 Sep '14

21 Sep '14

What is the best way to implement data structures that use pointers in Haskell? My first thought is as follows. Suppose for simplicity the labels are just integers, and each object need to point to one other one. I would have a getObject:: Int -> a f:: Int -> Int But I don't understand how Haskell maintains functions in memory, and in particular whether lookup/update is efficient. So suppose I define a f f x = case x of 1 -> 2 2 -> 3 .... (etc.) Q: if there are n entries here, how long does looking up (f m) take? Suppose later I update, with something like this: update::Int->Int->(Int->Int)->(Int->Int) update x y f z = (if z==x then y else f x) Q: how is "update x y f" maintained? Is using a f::Int->Int like using a pointer in terms of time complexity, and if not, what data structure should I use instead? Thanks, Holden

3 2