July 2011 - Haskell-Cafe

Jenkins/Hudson support for GHC warnings?
by Bryan O'Sullivan 07 Jul '11

07 Jul '11

Hi, folks - I know there are quite a few Haskell projects using the Jenkins (formerly Hudson) continuous build system, and I wonder if anyone has figured out how to get it reliably collecting errors and warnings from GHC. The standard warnings plugin<https://wiki.jenkins-ci.org/display/JENKINS/Warnings+Plugin>does a bad job - it tends to identify random (sometimes even empty) bits of build output as warnings. If you've configured it to behave better with GHC / hpc / QuickCheck / cabal-install, you'd be doing us a great service by sharing your tricks! Thanks, Bryan.

1 0

how to change a process name
by Kazu Yamamoto 07 Jul '11

07 Jul '11

Hello, I would like to know how to change a process name in Haskell. When we are programming in C, we can change it by overriding argv on Unix. But I cannot find the same way to do in Haskell. Can anyone suggest how in Haskell? I'm not talking about the result of System.Environment.getProgName but talking about the process name which we can see by the "ps" command. Thanks. --Kazu

2 1

Web Framework to Show Experimental Data
by Dmitri O.Kondratiev 07 Jul '11

07 Jul '11

Hi! I am looking for the Haskell libraries to quickly create some Web Framework to Show Experimental Data. Please share your experience and ideas. What tools would you chose for such a framework: == Architecture There are several unsynchronized processes: 1) Finder - demon that periodically searches and collects JSON data from the Web and stores this data in some sort of Archive (DB). 2) Processor - demon that periodically process new data from Archive collected by Finder and stores results of its work in the same Archive. 3) Web application that shows processed data in a browser. == General Requirements *** Main requirement - simplicity and high speed of development *** - Max Data storage capacity - 10 Mb - Max number of Web users - 50 - Web UI shows what Archive has at the moment of user request. == Web UI Requirements - Multi-page view of long data sets (some sort of scrolling) - Simple graphs and charts (desirable) == Questions 1) What to use for the Archive? Though input data is in JSON format, data generated from it is binary vectors. Not sure that CouchDB is a good choice in this case. What simple (Haskell lib + DB) combination one may advise that support concurrent read / write access with simple locking (don't need full transactional support)? 2) Web framework? Don't need AJAX, just basic stuff but simple and easy to code forms? Thanks!

2 3

CFP: PADL'12 - Practical Aspects of Declarative Languages 2012
by Claudio Russo 07 Jul '11

07 Jul '11

[Apologies if you receive multiple copies.] Call for Papers =============== 14th International Symposium on Practical Aspects of Declarative Languages (PADL 2012) http://research.microsoft.com/~crusso/padl12 Philadelphia, Pennsylvania, USA, January 23-24, 2012 Co-located with ACM POPL'12 Abstract/Paper submission deadline: September 10th/17th, 2011 Conference Description ====================== Declarative languages build on sound theoretical bases to provide attractive frameworks for application development. These languages have been successfully applied to many different real-world situations, ranging from data base management to active networks to software engineering to decision support systems. New developments in theory and implementation have opened up new application areas. At the same time, applications of declarative languages to novel problems raise numerous interesting research issues. Well-known questions include designing for scalability, language extensions for application deployment, and programming environments. Thus, applications drive the progress in the theory and implementation of declarative systems, and benefit from this progress as well. PADL is a forum for researchers and practitioners to present original work emphasizing novel applications and implementation techniques for all forms of declarative concepts, including, functional, logic, constraints, etc. Topics of interest include, but are not limited to: * Innovative applications of declarative languages * Declarative domain-specific languages and applications * Practical applications of theoretical results * New language developments and their impact on applications * Declarative languages and Software Engineering * Evaluation of implementation techniques on practical applications * Practical experiences and industrial applications * Novel uses of declarative languages in the classroom * Practical extensions such as constraint-based, probabilistic, and reactive languages. PADL'12 welcomes new ideas and approaches pertaining to applications and implementation of declarative languages. In this occasion PADL is co-located, as traditionally, with ACM POPL, which will be held immediately following PADL, January 25-27. The symposium will be held in Philadelphia, Pennsylvania, USA. Important Dates and Submission Guidelines ========================================= Abstract Submission: September 10, 2011 Paper Submission: September 17, 2011 Notification: October 22, 2011 Camera-ready: November 5, 2011 Symposium: January 23-24, 2012 Authors should submit an electronic copy of the full paper in PDF using the Springer LNCS format. The submission will be done through EasyChair conference system. If electronic submission is impossible, please contact the program chairs for information on how to submit hard copies. All submissions must be original work written in English. Submissions must be unpublished and not submitted for publication elsewhere. Work that already appeared in unpublished or informally published workshops proceedings may be submitted. PADL'12 will accept both technical and application papers: * Technical papers must describe original, previously unpublished research results. Technical papers must not exceed 15 pages in Springer LNCS format. * Application papers are a mechanism to present important practical applications of declarative languages that occur in industry or in areas of research other than Computer Science. Application papers will be published in the Springer-Verlag conference proceedings, and will be presented in a separate session. Application papers are expected to describe complex and/or real-world applications that rely on an innovative use of declarative languages. Application descriptions, engineering solutions and real-world experiences (both positive and negative) are solicited. The limit for application papers is 6 pages in Springer LNCS format. Program Committee ================= Marcello Balduccini, Intelligent Systems Department, Kodak Research Labs Edwin Brady, University of St Andrews, Scotland Henning Christiansen, Roskilde University, Denmark Agostino Dovier, University of Udine, Italy Matthew Flatt, University of Utah, USA Gopal Gupta, University of Texas at Dallas, USA John Hughes, Chalmers University of Technology, Sweden; Quviq AB Gabriele Keller, University of New South Wales, Australia Lunjin Lu, Oakland University, USA Marc Pouzet, École normale supérieure, France Ricardo Rocha, University of Porto, Portugal Andreas Rossberg, Google Germany GmbH, Germany Claudio Russo, Microsoft Research Cambridge, UK (co-chair) Kostis Sagonas, Uppsala Univeristy, Sweden Satnam Singh, Microsoft Research Cambridge, UK Zoltan Somogyi, University of Melbourne, Australia Eijiro Sumii, Tohoku University, Japan Terrance Swift, Universidade Nova de Lisboa, Portugal; Johns Hopkins University, USA Andrew Tolmach, Portland State University, USA Jan Wielemaker, University of Amsterdam, The Netherlands Roland Yap, National University of Singapore, Republic of Singapore Kwangkeun Yi, Seoul National University, Korea Neng-Fa Zhou, Brooklyn College, City University of New York, USA (co-chair) Contacts ======== For additional information about papers and submissions, please contact the Program Chairs: Claudio Russo Microsoft Research Cambridge,UK Email: crusso <AT> microsoft <DOT> com Neng-Fa Zhou Brooklyn College, The City University of New York, USA Email: zhou <AT> sci <DOT> brooklyn <DOT> cuny <DOT> edu With the Cooperation of ======================= The Association for Logic Programming (ALP) ACM SIGPLAN ===================================

1 0

Installation failure crypto-api on MAC
by Neil Davies 07 Jul '11

07 Jul '11

Hi Anyone out there seen this - building > Building crypto-api-0.6.3... > [2 of 9] Compiling Crypto.CPoly ( Crypto/CPoly.hs, dist/build/Crypto/CPoly.o ) several minutes later > ghc: could not execute: /usr/bin/gcc looks like it is silently running out of memory (I understand that this particular module is tough for the compiler due to the large 'constant' table inside it) I'm using "The Glorious Glasgow Haskell Compilation System, version 7.0.3" on (uname -a follows) Darwin mac-server-a.local 10.7.0 Darwin Kernel Version 10.7.0: Sat Jan 29 15:17:16 PST 2011; root:xnu-1504.9.37~1/RELEASE_I386 i386 Suggestions anyone? Neil

1 0

Re: [Haskell-cafe] NLP libraries and tools?
by wren ng thornton 07 Jul '11

07 Jul '11

On 7/6/11 5:58 PM, Aleksandar Dimitrov wrote: > On Wed, Jul 06, 2011 at 09:32:27AM -0700, wren ng thornton wrote: >> On 7/6/11 9:27 AM, Dmitri O.Kondratiev wrote: >>> Hi, >>> Continuing my search of Haskell NLP tools and libs, I wonder if the >>> following Haskell libraries exist (googling them does not help): >>> 1) End of Sentence (EOS) Detection. Break text into a collection of >>> meaningful sentences. >> >> Depending on how you mean, this is either fairly trivial (for English) or >> an ill-defined problem. For things like determining whether the "." >> character is intended as a full stop vs part of an abbreviation; that's >> trivial. > > I disagree. It's not exactly trivial in the sense that it is solved. It is > trivial in the sense that, usually, one would use a list of know abbreviations > and just compare. This, however, just says that the most common approach is > trivial, not that the problem is. Perhaps. I recall David Yarowsky suggesting it was considered solved (for English, as I qualified earlier). The solution I use is just to look at a window around the point and run a standard feature-based machine learning algorithm over it[1]. Memorizing known abbreviations is actually quite fragile, for reasons you mention. This approach will give you accuracy in the high 90s, though I forget the exact numbers. [1] With obvious features like whether the following word is capitalized, whether the preceding word is capitalized, length of the preceding word, whether there's another period after the next word,... >> But for general sentence breaking, how do you intend to deal with >> quotations? What about when news articles quote someone uttering a few >> sentences before the end-quote marker? So far as I'm aware, there's no >> satisfactory definition of what the solution should be in all reasonable >> cases. A "sentence" isn't really very well-defined in practice. > > As long as you have one routine and stick to it, you don't need a formal > definition every linguist will agree on. Computational Linguists (and their > tools,) more often than not, just need a dependable solution, not a correct one. But the problem is that what constitutes an appropriate solution for computational needs is still very ill-defined. For example, the treatment of quotations will depend on the grammar theory used in the tagger, parser, translator, etc. The quality of output is often quite susceptible to EOS being meaningfully[2] distributed. Thus, what constitutes a "dependable" solution often varies depending on the task in question.[3] Also, a lot of the tools in this area assume there's some sort of punctuation marking the end of sentences, even if it's unreliable as an EOS indicator. That works well enough for languages with European-like orthographic traditions, but it falls apart quite rapidly when moving to East Asian languages (e.g., Burmese, Thai,...). And languages like Japanese or Arabic can have "sentences" that go on forever, but are best handled by chunking them into clauses. [2] In a statistical sense, relative to the structure of the model. [3] Personally, I think the idea of having a single EOS type is the bulk of the problem. If we allowed for different kinds of EOS in grammars then the upstream tools could handle sentence fragments better, which would make it easier to make fragment breaking reliable. >> I've been working over the last year+ on an optimized HMM-based POS >> tagger/supertagger with online tagging and anytime n-best tagging. I'm >> planning to release it this summer (i.e., by the end of August), though >> there are a few things I'd like to polish up before doing so. In >> particular, I want to make the package less monolithic. When I release it >> I'll make announcements here and on the nlp@ list. > > I'm very interested in your progress! Keep us posted :-) Will do :) -- Live well, ~wren

2 1

Re: [Haskell-cafe] NLP libraries and tools?
by wren ng thornton 07 Jul '11

07 Jul '11

On 7/6/11 6:45 PM, Aleksandar Dimitrov wrote: > One hint, if you ever find yourself reading in quantitative linguistic data with > Haskell: forget lazy IO. Forget strict IO, except your documents aren't ever > bigger than a few hundred megs. In case you're not keeping the whole document in > memory, but you're keeping some stuff in memory, never keep it around in > ByteStrings, but use Text or SmallString (ByteStrings will invariably leak space > in this scenario.) Learn how to use Iteratees and use them judiciously. I definitely agree with the iteratees comment, but I'm curious about the leaks you mention. I haven't run into leakiness issues (that I'm aware of) in my use of ByteStrings for NLP. -- Live well, ~wren

2 1

GHC/ARM registerised port
by Karel Gardas 07 Jul '11

07 Jul '11

Hello, Stephen Blackheath, David Terei and me are working together on ARM registerised port of GHC. The port is using LLVM as a code generator and is kind of working already. (GHCi support still missing) If you are curious and would like to try the code, please read last two paragraphs containing instructions what to do here[1]. This blog[2] is my personal attempt to document the project at least from my point of view. If you would just like to go straight to the code for code review or to follow progress directly, then github.com repo[3] is the right link for you. Thanks for testing, bug reporting and/or patches! Karel [1]: http://ghcarm.wordpress.com/2011/07/06/armv7-thumb-vfpv3-support-and-github… [2]: http://ghcarm.wordpress.com/ [3]: https://github.com/kgardas/ghc

1 0

[Haskell-Cafe] Constructive Probability Theory and RandProc's σ-algebras
by Edward Kmett 07 Jul '11

07 Jul '11

On Sun, Jul 3, 2011 at 11:05 AM, David Banas <dbanas(a)banasfamily.net> wrote: > v0.4 of `RandProc` has just been posted to Hackage: > http://hackage.haskell.org/package/randproc [NB: I transfered this discussion from haskell to haskell-cafe, and started another thread] I have been spending some time exploring constructive probability theory in Haskell, and I have a couple of observations about working with random processes in a constructive setting: Necessarily because you are working in Haskell, a constructive setting, your σ-algebras can't be actual σ-algebras! =/ Notably, they aren't closed under *countable* union/intersection/complement, but merely finite union/intersection/complement, and Gray and Davisson's horrified objections from the bottom of page 39 on Random Processes applies: If a class of sets is only a field rather than a σ-field, that is, if it > satisfies only (3.1) and (3.2), then there is no guarantee that the class > will contain all limits of sets. Hence, for example knowing that a class of > sets contains all half-open intervals of the form (a,b] for a and b finite > does not ensure that it will also contain points or singleton sets! In fact, > it is straightforward to show that the collection of all such half-open > intervals together with the complements of such sets and all finite unions > of the intervals and complements forms a field. The singleton sets, however, > are not in the field. (See exercise 3.4) > Thus if we tried to construct a probability theory based on only a field, > we might have probabilites defined for events such as (a,b) meaning "the > output voltage of a measurement is between a and b" yet not have > probabilities defined for a singleton set {a} meaning "the output voltage is > exactly a." By requiring that the event space be a σ-field instead of only a > field, we are assured that all such limits are indeed events. This limitation, while from some perspective annoying is intrinsic to tackling the problem intuitionistically. Practically, this means you need to be careful when rederiving most of the later results, because you are limited to what you can prove in a constructive measure theory. In particular the approaches of YK Chan, Brian Weatherson, and Glenn Shafer to intuitionistic probability theory are probably appropriate reading. Personally, I don't think this is all that bad, we get some nice properties by being in a constructive setting. Weatherson noted that your measure becomes additive without problems from Dutch book arguments in an intuitionistic setting, because P(A v not A) is not necessarily 1! You may recall that recall that additivity would imply that P(A v B) = P(A) + P(B), given that A and B are disjoint, but that it tends to fall apart in classical probability theory. Intuitionistically, however, this is fine. That is to say that if you placed bets that payed out with market rate interest at a rational booking agent on both whether P = NP and P /= NP, it isn't just as good as having placed a bet on truth, because intuitionistically it is possible that neither of those bets may ever pay out, but in a classical setting P (P = NP v P /= NP) = P(True) = 1, so we lose additivity to gain the excluded middle. The cost of additivity is giving up or refining a lot of results from that text that talk about the limit of a random process. That and that the "σ-fields" in question are mere fields. ;) I am interested in exploring the consequences of this further. -Edward

2 1

Re: [Haskell-cafe] NLP libraries and tools?
by wren ng thornton 07 Jul '11

07 Jul '11

On 7/6/11 8:46 PM, Richard O'Keefe wrote: >> I've been working over the last year+ on an optimized HMM-based POS >> tagger/supertagger with online tagging and anytime n-best tagging. I'm >> planning to release it this summer (i.e., by the end of August), though >> there are a few things I'd like to polish up before doing so. In >> particular, I want to make the package less monolithic. When I release it >> I'll make announcements here and on the nlp@ list. > > One of the issues I've had with a POS tagger I've been using is that it > makes some really stupid decisions which can be patched up with a few > simple rules, but since it's distributed as a .jar file I cannot add > those rules. How horrid. I assume the problem is really that the trained model is in the jar and you can't do your own training? Or is this a Brill-like tagger where you really mean to add new rules? If an HMM-based tagger is amenable, you could try switching to Daniël de Kok's Java port of TnT: https://github.com/danieldk/jitar The tagger I'm working on does support being hooked up to a Java client (i.e., consumer of tagging info), but it's fairly ugly due to Java's refusal to believe in IPC. -- Live well, ~wren

1 0