- Haskell-Cafe - Haskell.org

Proposal: move ansi-terminal to haskell-pkg-janitors
by Roman Cheplyaka 17 Dec '13

17 Dec '13

ansi-terminal is a useful and important package providing colorful console output. It has 50 direct reverse dependencies, and a huge number of indirect ones (through testing frameworks, for example). ansi-terminal was written and maintained by Max Bolingbroke, but he doesn't seem to be active in the Haskell community anymore. The package itself hasn't been updated for a year. Unless he (or someone else) objects, I propose to move ansi-terminal under maintainership of the haskell-pkg-janitors group. Is there any venue for haskell-pkg-janitors, such as a mailing list or IRC channel? Roman

3 5

text 1.0.0.0 producing cabal hell?
by Tillmann Vogt 17 Dec '13

17 Dec '13

Check this out: http://packdeps.haskellers.com/reverse/text 306 outdated dependencies! I am guessing that most libraries could easily change the boundaries to >=0.11 && <1.1 <http://hackage.haskell.org/package/digestive-functors-blaze> because the API hasn't changed a lot. As far as I can see only a function was added. Or is everybody using sandboxing tools? Maybe the author of a library should have the right to change the boundary of all reverse dependencies. So that Brian has to adjust all 306 boundaries if he knows that this would work :-) Just the boundaries, not the code. -Tillmann

4 3

Re: [Haskell-cafe] How to determine the right path to haddock html documentation?
by Roman Cheplyaka 17 Dec '13

17 Dec '13

* Carlo Hamalainen <carlo(a)carlo-hamalainen.net> [2013-12-17 13:10:43+0100] > On 17/12/13 11:32, Roman Cheplyaka wrote: > > > > First, no need to use that fork — all patches have been merged into > > Cabal master at https://github.com/haskell/Cabal. (However, they are not > > released yet, so you still have to build the git version.) > > OK, I've built cabal from github in a clean test account. > > > Second, no, installation won't happen automatically. At the moment you > > have to do that manually, like this: > > > > cabal install --haskell-suite -w hs-gen-iface mtl > > I may be missing something obvious, but which package provides > haskell-suite? I see that it's a group of repositories on github. haskell-suite is not a real package or program. In this context, it's the name of a "virtual compiler" known to cabal. > $ cabal install --haskell-suite -w parsec > cabal: The program haskell-suite is required but it could not be found. The correct command is $ cabal install --haskell-suite -w hs-gen-iface parsec The -w flag lets cabal know that you want to compile using hs-gen-iface, which is a concrete compiler implementing the haskell-suite interface. > > It'd be nice to add a cabal.config option to do that automatically. > > Ditto for --haddock-hyperlink-source which I use all the time so that I > get locally built documentation. Very handy for when I work at a cafe. Yeah, ideally haddock and hscolor should be just separate compilers, just like ghc and hs-gen-iface are. The user would then specify in cabal.config what set of compilers to use for simple `cabal install $pkg`. > I want to make the development process in Vim as smooth as possible. A > newbie should be able to ask "what's this?" or "where is this from?" and > get a quick answer, right there in the editor. Hoogle is very useful but > there should be an editor-integrated solution that works locally and > gives an answer like "this symbol in this file comes from this module on > this system". Then you may be also interested in Ariadne :) https://github.com/feuerbach/ariadne Roman

1 0

Frankfurt Haskell User Group Meetup - Wed. 18.12.
by Peter Althainz 17 Dec '13

17 Dec '13

Dear All, The next meeting of Frankfurt Haskell meetup takes place on Wednesday, 18th of Dec. and is about programming a game with HGamer3D. See game video here: http://www.youtube.com/watch?v=DKq0YsWRPoc and link to invitation here: http://www.meetup.com/Frankfurt-Haskell-User-Group/events/150475792/. Also we will have time for hacking on HWebUI. Peter

1 0

ANNOUNCE: New OpenAL packages
by Sven Panne 17 Dec '13

17 Dec '13

New versions of the OpenAL packages are available on Hackage: * OpenAL 1.5.0.0 * ALUT 2.3.0.0 The main change is that the installation doesn't require autoconf anymore, instead of that, a pure Cabal-based build system is used now. This should make installing the packages on non-Linux systems much easier. Apart from that, uniqueBuffers works now and iOS is a supported platform, thanks to Stephen Blackheath. The packages depend on OpenGL now instead of ObjectName/StateVar/Tensor, because the latter packages were assimilated into OpenGL and there are probably very few programs using OpenAL without OpenGL. Have fun, S.

1 0

haddock for all modules (exposed and other) in a cabalized project?
by Johannes Waldmann 17 Dec '13

17 Dec '13

I have a cabalized project with few "exposed-modules", and a lot of "other-modules". How can I invoke "cabal haddock --hyperlink-source" for *all* modules? - Thanks, J.W.

1 0

Cabal sandbox status in your ZSH prompt
by Daniil Frumin 17 Dec '13

17 Dec '13

Hi everyone, I made a simple script for my zsh setup that allows me to see whether am I in a cabalized sandbox environment or not. On my machine it looks like this: <https://files.app.net/rjphjAG9.png> The script itself is here: <https://gist.github.com/co-dan/7963200> The result of checking for the sandbox is cached, which---as I've realized only moments ago---is probably unnecessary; it updates only when the user performs a `cabal` command or changes a directory. -- Sincerely yours, -- Daniil Frumin

3 2

Fwd: Polymorphic updating with TC/TFs?
by adam vogt 17 Dec '13

17 Dec '13

Hello Hans, You can move the setFoo into a multiparameter type class, to which you can add that constraint: class (HasFoo a, HasFoo b, b ~ NoFoo a (Foo b)) => SetFoo a b where setFoo :: Foo b -> a -> b instance (HasFoo b, b ~ NoFoo (FooT f a) (Foo b)) => SetFoo (FooT f a) b where setFoo f (FooT _ x) = FooT f x In my opinion, that's not as nice as the fundep solution which has less names: class Has s t a b | s -> a, t -> b, s b -> t, t a -> s On the topic of defining setFoo in terms of setFoo', you might be interested in -XDefaultSignatures, which can work with the SetFoo class defined above: class HasFoo a where default setFoo' :: SetFoo a a => Foo a -> a -> a setFoo' = setFoo setFoo' :: Foo a -> a -> a Finally, there seems to be some overlap with generic programming. Changing type parameters isn't supported by the ones I know (syb), but you can still fake it: <http://www.haskell.org/haskellwiki/SYB#fmap>. Regards, Adam On Mon, Dec 16, 2013 at 7:37 PM, Hans Höglund <hans(a)hanshoglund.se> wrote: > Hello, > > I am working with a set of type classes of the following form (http://lpaste.net/97110) The idea is that every such class provide an associated type Foo, and a lens to the Foo in every instance. I.e. this class is used to provide view/set/modify for all types that contain a Foo somewhere deep in its structure. > > For "simple lenses", i.e. functions that does not modify the associated Foo, this is straightforward. However to support polymorphic updates it seems necessary to add another associated type NoFoo, which must be used to constraint the return type of set. > > What bothers me is the redundancy of the two set functions. I would intuitively expect set' to be implemented in terms of set (as it seems to be a restriction of that function), but this is not possible, as the compiler can not deduce that (NoFoo a (Foo a) ~ a). Is there a way to add this constraint to the type class? > > Hans > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe

1 0

Building applications with profiling
by Mateusz Kowalczyk 17 Dec '13

17 Dec '13

Greetings, I want to profile the Yi text editor. A quick look on the Haskell wiki suggests that I need --enable-executable-profiling, however enabling that complains that the yi libraries weren't built with profiling. Additionally enabling --enable-library-profiling allows me to recompile the libs but at the very end I get an error about ‘mtl’ not having been built with profiling. >src/library/Yi/Boot.hs:9:8: > Could not find module `Control.Monad.State' > Perhaps you haven't installed the profiling libraries for package >`mtl-2.1.2'? > Use -v to see a list of the files searched for. I can't really recompile mtl with profiling considering that just about every single package on my system seems to depend on it and they would (probably?) all break. Does anyone have any suggestions on how to approach the problem? Am I doing it all wrong? Thanks -- Mateusz K.

2 3

Polymorphic updating with TC/TFs?
by Hans Höglund 17 Dec '13

17 Dec '13

Hello, I am working with a set of type classes of the following form (http://lpaste.net/97110) The idea is that every such class provide an associated type Foo, and a lens to the Foo in every instance. I.e. this class is used to provide view/set/modify for all types that contain a Foo somewhere deep in its structure. For "simple lenses", i.e. functions that does not modify the associated Foo, this is straightforward. However to support polymorphic updates it seems necessary to add another associated type NoFoo, which must be used to constraint the return type of set. What bothers me is the redundancy of the two set functions. I would intuitively expect set' to be implemented in terms of set (as it seems to be a restriction of that function), but this is not possible, as the compiler can not deduce that (NoFoo a (Foo a) ~ a). Is there a way to add this constraint to the type class? Hans

1 0