- Haskell-Cafe - Haskell.org

ghc on xeon phi?
by Dominic Steinitz 22 Feb '17

22 Feb '17

> Hi Levent, > >> The whole point of the Xeon-phi is the availability of >> large-vector sized floating-point support and many-many cores. > > Sure, that's what I'm contemplating - use the many options > of writing parallel and concurrent Haskell programs. > > So, GHC's RTS should "just work"? > > I was hoping someone already had actually seen this > on their machine. > > - J. I recall we thought about using a Xeon phi for our application but abandoned the attempt for reasons I cannot recall. I’ll see if I can find out. But maybe you would better using something like accelerate-llvm? I think you wrote your application was numerical. BTW I am currently adding a small improvement to ghc floating point code generation but I am beginning to think it’s going to be difficult to get ghc to generate really good numerical / floating point code and I am experimenting with accelerate-llvm. Dominic

1 0

A question about GHC test coverage
by Olaf Klinke 22 Feb '17

22 Feb '17

You should not have to write tests for functions you did not define. Correct me if I'm wrong, but any property of max can be derived from the properties of <=. Sadly, the laws for Ord are not stated in the documentation of Data.Ord. (Because there is no consensus?) They are: Reflexive: if x == y then x <= y Transitive: if x <= y and y <= z then x <= z Antisymmetric: if x <= y and y <= x then x == y Olaf

2 3

What is this applicative functor?
by Joachim Breitner 20 Feb '17

20 Feb '17

Hi, I recently wrote this applicative functor: data OneStep a = OneStep a [a] instance Functor OneStep where fmap f (OneStep o s) = OneStep (f o) (map f s) instance Applicative OneStep where pure x = OneStep x [] OneStep f fs <*> OneStep x xs = OneStep (f x) (map ($x) fs ++ map f xs) takeOneStep :: OneStep t -> [t] takeOneStep (OneStep _ xs) = xs This was useful in the context of writing a shrink for QuickCheck, as discussed at http://stackoverflow.com/a/41944525/946226. Now I wonder: Does this functor have a proper name? Does it already exist in the libraries somewhere? Should it? Greetings, Joachim -- Joachim “nomeata” Breitner mail(a)joachim-breitner.de • https://www.joachim-breitner.de/ XMPP: nomeata(a)joachim-breitner.de • OpenPGP-Key: 0xF0FBF51F Debian Developer: nomeata(a)debian.org

8 14

In case there are some other Haskellers around Eindhoven, NL
by Damian Nadales 20 Feb '17

20 Feb '17

On a very unmemorable day, lost somewhere in a province of the Netherlands, a new Haskell meetup has been born http://meetu.ps/c/32F0Z/tJxcH/f Best, Damian.

3 2

Selective deriving
by Rui Azevedo 20 Feb '17

20 Feb '17

Hi! I don't know if there is some extension that already does this, or any other method of doing it. My search on deriving ghc mechanisms and extensions gave me no satisfactory results. the point is to have some "help" on derivation or some sort of derivation. actual derivation schema seems to require that all members (constructor arguments) instantiate the deriving class. This is the way of doing it for classes like Eq, works for Show and Ord but eventually it is a very strong requirement even for Eq sometimes. Of course I can do a manual derivation to state something else, and that sometimes can be frustrating because the derivations are way trivial and susceptible of automation. There is no fancy things added, its all standard haskell, user could laboriously type this. As an alternate derivation mechanism i seek selective deriving, like this: class Named o where name::o->String data A = A String deriving (Show) instance Named A where name="Object" till now just normal derivation data B = B A Int deriving (Show) extending (Named) the mechanism on extending requires that one and only one of the type constructor's direct arguments instantiates the class. this avoids the C++ diamond inheritance problem, user can recur on manual derivation. this will generate the code: instance Named B where name (B a _)=name a alternative can be: extending (Named A,...) this latter form is rater interesting as it cuts down the implementation boilerplate and still avoid the diamond problem even if multiple members instantiate the same class Are already any extension doing this? Can i achieve the same result by other forms? What is the best way to implement this thing? plug-ins seem not to be able to change syntax (right?) I've also looked at -F -pgmF options to use a preprocessor... can a preprocessor insert code to reference the original line numbers, just because messing with line numbers will make this unusable. regards rui

2 1

AMMCS-2017 call for submissions
by Jacques Carette 19 Feb '17

19 Feb '17

Applied Mathematics, Modeling and Computational Science (AMMCS-2017) Waterloo, Ontario, Canada August 20-25, 2017 Minisymposium: Symbolic Methods in Code Generation and Programming Languages (ID: SS-SMCGPL; http://www.ammcs2017.wlu.ca/special-sessions/smcgpl/) Symbolic Methods in Code Generation and Programming Languages Symbolic methods are increasingly finding a wide variety of applications in and around programming languages. Starting from their immensely successful use in generation of numerical codes, their applicability is widening to both code analysis (often through abstract interpretation) and even in certain phases of compilers themselves, as exemplified in recent probabilistic programming languages. This session will bring together experts in symbolic computation along with programming languages experts to exchange ideas on the state of the art in this domain. Abstracts are being accepted for special sessions, contributed sessions, and poster presentations. All participants should submit their abstracts online via the AMMCS-2017 Management System at http://www.ammcs2017.wlu.ca/submit-abstracts/ Due: March 15, 2017 SS-SMCGPL Organizers: Jacques Carette (McMaster University), Eugene Zima (Wilfrid Laurier University) About AMMCS =========== AMMCS-2017 is an interdisciplinary international conference in a series of AMMCS meetings held in Waterloo, Ontario, Canada. The AMMCS Conference Series aims at promoting interdisciplinary research and collaboration involving mathematical and computational sciences within a larger international community, and highlighting recent advances in Applied Mathematics, Modeling and Computational Science (AMMCS). In 2017 the conference will be held in August (August 20-25, 2017). From its inception this conference series has been organized in cooperation with AIMS and SIAM. The conference provides a unique opportunity for in-depth technical discussions and exchange of ideas in all areas involving mathematical and computational sciences, modeling and simulation, as well as their applications in natural and social sciences, engineering and technology, industry and finance. It offers to researchers, industrialists, engineers and students to present their latest research, to interact with the experts in the field, and to foster interdisciplinary collaborations required to meet the challenges of modern science, technology, and society.

1 0

QuickCheck and State monad
by George Boulougaris 19 Feb '17

19 Feb '17

Hello, I have written a Haskell module that contains functions that operate on some state. Let's say that the code looks like this (the actual functions may return an actual result instead of `()`, but this is irrelevant to the question): data StateContext = StateContext { -- some records } handleEventA :: EventA -> State StateContext () handleEventB :: EventB -> State StateContext () handleEventC :: EventC -> State StateContext () As you can imagine the behavior of each function depends on the current state. For example `handleEventA >> handleEventB` will not produce the same result as `handleEventB >> handleEventA`. So I have several HUnit tests that verify the behavior of each function at several states. But now I would like to write more tests, that exercise all functions at all possible states (the number of states is finite). Writing them with HUnit would be quite labor-itensive, so I thought that using QuickCheck might be helpful in that case (I have only used it before for trivial functions). But I cannot see which properties should I test, or what kind of data should the test generate. I suspect that the test should generate random sequence of events (e.g. `handleEventB >> handleEventC >> handleEventA` etc), but I cannot see what properties should be satisfied. Thanks

3 3

[ANN] marvin-0.1.0, a Haskell framework for writing chat bots
by Justus Adam 19 Feb '17

19 Feb '17

Hello Haskell enthusiasts, For a couple of months now I’ve been working on a framework for writing chat bots in Haskell. The project is very heavily inspired by Hubot (https://hubot.github.com/ <https://hubot.github.com/>) and for the most part mimics its API. However it provides all the nice Haskell features, such as type safety and running as compiled code, rather than interpreted. If you’re curious check it out, I’d love some feedback. (Verbose) Documentation: http://marvin.readthedocs.io <http://marvin.readthedocs.io/> Package: https://hackage.haskell.org/package/marvin <https://hackage.haskell.org/package/marvin> Source code repository: https://github.com/JustusAdam/marvin <https://github.com/JustusAdam/marvin> Suggestions and issues: https://github.com/JustusAdam/marvin/issues <https://github.com/JustusAdam/marvin/issues> Regards, Justus

2 3

ThEdu'17: Call for Extended Abstracts & Demonstrations
by Achim D. Brucker 18 Feb '17

18 Feb '17

(Apologies for duplicates) Call for Extended Abstracts & Demonstrations ThEdu'17 Theorem proving components for Educational software http://www.uc.pt/en/congressos/thedu/thedu17 at CADE 26 International Conference on Automated Deduction 6-11 August 2017 Gothenburg, Sweden http://www.cade-26.info/ ThEdu'17 Scope Computer Theorem Proving is becoming a paradigm as well as a technological base for a new generation of educational software in science, technology, engineering and mathematics. The workshop brings together experts in automated deduction with experts in education in order to further clarify the shape of the new software generation and to discuss existing systems. Topics of interest include: * methods of automated deduction applied to checking students' input; * methods of automated deduction applied to prove post-conditions for particular problem solutions; * combinations of deduction and computation enabling systems to propose next steps; * automated provers specific for dynamic geometry systems; * proof and proving in mathematics education. Important Dates Extended Abstracts: 18 June 2017 Author Notification: 2 July 2017 Final Version: 16 July 2017 Workshop Day: 6 August 2017 Submission Interested researchers are invited to submit extended abstracts and system descriptions. Both kinds of submissions should be approximately 5 pages in length and present original unpublished work not submitted elsewhere. Submission is in PDF format via easychair, https://easychair.org/conferences/?conf=thedu17 formatted according to http://www.easychair.org/publications/easychair.zip The extended abstracts and system descriptions will be made available online. At least one author is expected to presents his/her submission at ThEdu'17. Joint publication in companion with other CADE26 events is under consideration (as a volume in the EPiC Series in Computing). Program Committee Francisco Botana, University of Vigo at Pontevedra, Spain Achim D. Brucker, University of Sheffield, United Kingdom Filip Maric, University of Belgrade, Serbia Walther Neuper, Graz University of Technology, Austria (co-chair) Pavel Pech , University of South Bohemia, Czech Republic Pedro Quaresma, University of Coimbra, Portugal (co-chair) Vanda Santos, CISUC, Portugal Wolfgang Schreiner, Johannes Kepler University, Austria Burkhart Wolff, University Paris-Sud, France -- Dr. Achim D. Brucker | Software Assurance & Security | University of Sheffield https://www.brucker.ch | https://logicalhacking.com/blog @adbrucker | @logicalhacking

1 0

Getting a rewrite rule to fire
by Clinton Mead 18 Feb '17

18 Feb '17

Basically, I want to rewrite `g (f x)` with `h x` where it's valid to do so (i.e. appropriate instances of `h` exist). The code I've put below is a silly example just to illustrate the issue. I guess the tricky thing is that whether the rewrite rule can fire depends on the result of `g (f x)`, not just x itself. Does anyone know how to adjust this so the rewrite rule fires? {-# LANGUAGE TypeFamilies #-} {-# LANGUAGE TypeApplications #-} module Main where data D a {-# INLINE [1] f #-} f :: a -> D a f = undefined type family T a type instance T Char = Int {-# INLINE [1] g' #-} g' :: (G a) => D (T a) -> a g' = undefined class G a where g :: D (T a) -> a g = g' instance G Char class H a where h :: T a -> a main = ((g (f (2 :: Int))) :: Char) `seq` return () {-# RULES "myrule" forall (x :: H a => T a). g' (f x) = h @a x #-}

2 1