- Haskell-Cafe - Haskell.org

Space leaks in function that uses Data.Vector.Mutable
by Andrey Yankin 23 Jan '13

23 Jan '13

Hi! I have a "low-level" function for insertion element into mutable vector. It looks like this: place :: (PrimMonad m) => MV.MVector (PrimState m) (Int, t) -> (Int, t) -> Int -> m () place v max@(val1,_) i = place' i where place' i = do let j = i - 1 if j < 0 then return () else do curr@(val2, _) <- MV.unsafeRead v j -- <<<<< if val2 > val1 then do MV.unsafeWrite v j max MV.unsafeWrite v i curr place' j else return () It searches a right place for the i-th element, moving it towards beginning . Surprisingly, it works, but is slow. Time profiling says this: COST CENTRE MODULE %time %alloc ticks bytes place.place' Main 40.1 77.1 9167 12169223232 And heap profiling says that most of allocations are for Int and (,). I think, binding of unsafeRead to curr and val might allocate my Precious memory more than, maybe, it is needed. Am I right? Is it some strictness that I need? Is there anything I can do in this situation? Or should I look outside of this function? PS Emergence of this piece of code is a long story. Originally I was trying to solve Project Euler's problem 411. But now it is a different issue.

2 1

ANN: hgdbmi 0.2, GDB Machine Interface
by Alexander Bernauer 23 Jan '13

23 Jan '13

Hi I am happy to announce the second release of hgdbmi, a Haskell implementation of the GDB Machine Interface. The Machine Interface is a machine-readable variant of the stdin/stdout-based interface of the GNU debugger. With hgdbmi you can control the execution of a GDB instance to drive debugging of arbitrary targets. Version 0.2 of this library is a substantial, non-compatible extension of version 0.1. The API provides a thread-safe and synchronous command-response function and supports user-defined call-backs for asynchronous notifications. All commands, responses and notifications are supported in this version. Also, for some response and notification types, we additionally provide semantic data types and corresponding translation functions. The package: http://hackage.haskell.org/package/hgdbmi-0.2 Greetings, Alex PS: This is my first serious Haskell package. If you see something that you consider odd, I might have done this simply because I didn't know better. I would appreciate if you would let me know in such a case.

2 4

The Coverage Condition of functional dependencies
by Petr P 23 Jan '13

23 Jan '13

Hi, trying to understand UndecidableInstances (and to find and answer to < http://stackoverflow.com/q/14476230/1333025>), I was trying to find out why "mtl" needs UndecidableInstances. The reason is that instances like > instance MonadState s m => MonadState s (ContT r m) where don't satisfy the Coverage Condition: "The Coverage Condition. For each functional dependency, tvsleft -> tvsright, of the class, every type variable in S(tvsright) must appear in S(tvsleft), where S is the substitution mapping each type variable in the class declaration to the corresponding type in the instance declaration. " (See http://www.haskell.org/ghc/docs/7.0.1/html/users_guide/type-class-extension… ) In other words, "s" isn't expressed using type variables in "ContT r m". But in these cases, it's actually possible. Because of the assertion "MonadState s m" and its dependency "m -> s" we know that "s" will be always deducible from "m". I wonder, would it be possible to augment the type checker to realize this? It seems reasonable: Before comparing if S(tvsright) is a subset of S(tvsleft), we'd add every type variable to S(tvsleft) that is determined from it using functional dependencies in the assertion of the instance. Best regards, Petr

1 0

Re: [Haskell-cafe] Mobile app development?
by Andrew Pennebaker 23 Jan '13

23 Jan '13

Can we un-deprecate GHC's ability to compile to C code? C may be the best option to bridge to mobile, as Android, iOS, and Windows RT do support C/C++ apps. On Jan 22, 2013 2:14 PM, "Dan Choi" <dhchoi(a)gmail.com> wrote: > > What about the option of using Haskell's Parsec or AttoParsec to implement > a Haskell-ish language that compiles down not to machine code but to > Objective C or Android Java? Like how CoffeeScript compiles down to > JavaScript. > > > > On Saturday, January 19, 2013 1:58:03 PM UTC-5, Casey Hawthorne wrote: >> >> You would need native compilers for all the platforms and/or virtual >> machine technology. >> >> Might be easier to have the browser connect to a Haskell app. >> >> >> On Sat, Jan 19, 2013 at 10:42 AM, Andrew Pennebaker >> <andrew.p...@gmail.com> wrote: >> > There are currently very few options, especially free and open source >> > options, when it comes to developing cross-platform mobile >> applications. >> > It's basically web apps with JavaScript, or C++. If Haskell supported >> app >> > development on Android, iOS, and Windows RT, that alone would bring in >> more >> > developers. >> > >> > Similarly, there are very few languages for mobile development that >> take >> > advantage of multiple cores and multiple CPUs. Haskell's `parmap` is an >> > amazing selling point. Can we please prioritize mobile support? I'd >> much >> > rather write everything in ML than PhoneGap. >> > >> > -- >> > Cheers, >> > >> > Andrew Pennebaker >> > www.yellosoft.us >> > >> > ______________________________**_________________ >> > Haskell-Cafe mailing list >> > Haskel...@haskell.org >> > http://www.haskell.org/**mailman/listinfo/haskell-cafe<http://www.haskell.org/mailman/listinfo/haskell-cafe> >> > >> >> >> >> -- >> -- >> Regards, >> KC >> >> ______________________________**_________________ >> Haskell-Cafe mailing list >> Haskel...@haskell.org >> http://www.haskell.org/**mailman/listinfo/haskell-cafe<http://www.haskell.org/mailman/listinfo/haskell-cafe> >> >

3 5

Re: [Haskell-cafe] ANN: Haskell bindings for the igraph C library
by George Giorgidze 22 Jan '13

22 Jan '13

Hi Jason, > How does this compare with fgl? http://hackage.haskell.org/package/fgl A couple of months ago, I was working on a project where I had to do some graph analysis. I encountered two problems with the fgl library: (1) The fgl library did not work for large graphs, even graph construction would fail with stack overflow. (2) The fgl library did not provide the graph algorithms needed in the project: closeness centrality, Brandes betweenness centrality, eigenvector centrality, pagerank and some more. The igraph library featured all functions that I needed and also worked for large graphs without problems. I was also very pleased with its performance and started writing Haskell bindings to igraph. Cheers, George

2 1

Re: [Haskell-cafe] [Haskell] ANNOUNCE: MFlow 0.2
by Alberto G. Corona 22 Jan '13

22 Jan '13

The template look is very simple but it uses a lot of dynamic code behind I changed the template to something more light. http://haskell-web.blogspot.com.es/ 2013/1/20 <briand(a)aracnet.com> > On Thu, 17 Jan 2013 16:07:41 +0100 > "Alberto G. Corona " <agocorona(a)gmail.com> wrote: > > > > > The entry in my blog, with the announcement and the philosophy behind > > http://haskell-web.blogspot.com.es/2013/01/announce-mflow-02.html > > > > FYI > That page never comes up for me. I just see spinning gears. > iceweasel 10.0.7 > > Brian > > -- Alberto.

2 1

Re: [Haskell-cafe] Mobile app development?
by KC 22 Jan '13

22 Jan '13

Then it looks as if the easier implementation would be small Haskell VM's for the various platforms with a byte code compiler. I do not believe the JVM supports all the optimizations GHC can do. Oh wait! Can the LLVM be easily ported to do this? On Sat, Jan 19, 2013 at 11:40 AM, Andrew Pennebaker <andrew.pennebaker(a)gmail.com> wrote: > >> Might be easier to have the browser connect to a Haskell app. > > > Not all apps can be run as thin clients. 3D video games and other intensive > programs aren't easily done as thin clients. Mobile Haskell would be very > powerful, because concurrency and parallelism aren't something the C-family > languages are supporting that well. > >> >> >> >> On Sat, Jan 19, 2013 at 10:42 AM, Andrew Pennebaker >> <andrew.pennebaker(a)gmail.com> wrote: >> > There are currently very few options, especially free and open source >> > options, when it comes to developing cross-platform mobile applications. >> > It's basically web apps with JavaScript, or C++. If Haskell supported >> > app >> > development on Android, iOS, and Windows RT, that alone would bring in >> > more >> > developers. >> > >> > Similarly, there are very few languages for mobile development that take >> > advantage of multiple cores and multiple CPUs. Haskell's `parmap` is an >> > amazing selling point. Can we please prioritize mobile support? I'd much >> > rather write everything in ML than PhoneGap. >> > >> > -- >> > Cheers, >> > >> > Andrew Pennebaker >> > www.yellosoft.us >> > >> > _______________________________________________ >> > Haskell-Cafe mailing list >> > Haskell-Cafe(a)haskell.org >> > http://www.haskell.org/mailman/listinfo/haskell-cafe >> > >> >> >> >> -- >> -- >> Regards, >> KC > > > > > -- > Cheers, > > Andrew Pennebaker > www.yellosoft.us -- -- Regards, KC

4 3

Non-derivable Typeable
by Roman Cheplyaka 22 Jan '13

22 Jan '13

I need to declare a Typeable instance for a type which has an argument of kind * -> *. GHC refuses to derive it. What is a recommended way to go about it? In particular, if I write the instance by hand, how important is the fingerprint, and how could I generate it? Roman

2 5

[ANN] tls-extra 0.6.1 - security update, please upgrade.
by Vincent Hanquez 22 Jan '13

22 Jan '13

Hi cafe, this is a security advisory for tls-extra < 0.6.1 which are all vulnerable to bad certificate validation. Some part of the certificate validation procedure were missing (relying on the work-in-progress x509 v3 extensions), and because of this anyone with a correct end-entity certificate can issue certificate for any arbitrary domain, i.e. acting as a CA. This problem has been fixed in tls-extra 0.6.1, and I advise everyone to upgrade as soon as possible. Despite a very serious flaw in the certificate validation, I'm happy that the code is seeing some audits, and would want to thanks Ertugrul Söylemez for the findings [1]. [1] https://github.com/vincenthz/hs-tls/issues/29 -- Vincent

3 5

regular expression sub-expression matching
by Nicolas Bock 21 Jan '13

21 Jan '13

Hello list, I was wondering whether there is a way to do what the following python code does in haskell: import re result = re.compile("M\(([0-9]+),([0-9]+)\) *= *([0-9.eE-]+);").search("M(1,2) = 0.1e-3;") if result: print(result.group(1), result.group(2), result.group(3)) Basically I would like to pattern match parts of a string and return the matches. I have looked at Text.Regex.Posix but (of course I might just not have understood the functions in there properly) it seems as if it does not provide anything like python's re module. Thanks already, nick

3 3