[Haskell-cafe] darcs patch dependencies in dot format
Hi all! Yesterday I wrote a little tool to output the dependencies of darcs patches in dot format. The hardest part was to wrap my head around the darcs API and find a way to let it compute the patch dependencies. I don't know, if I got it right, but it looks correct at first glance. Here is the code: https://patch-tag.com/r/shahn/darcs2dot To use it just execute the program in a darcs repo and it will output a dot graph to stdout. The graph does not contain all dependencies, but the transitive reduction. The patch names are truncated at 15 characters. And here is an example graph: http://open-projects.net/~shahn/patchDeps.pdf These are the patch dependencies of one of my darcs repos (https://patch-tag.com/r/shahn/hate). Some observations I made: * There are two completely separate subgraphs. One subgraph seems to be for the testsuite, the other for the actual code. This means, the two don't depend on each other and could easily be put in two distinct repos. * There is a re-implementation patch with a lot of incoming and outgoing edges. (Which makes sense.) * There are some sequences of patches (e.g. these four allow ... patches in the upper left corner) that seem to contain related patches. * darcs's patch system is awesome! Any comments or suggestions? Cheers, Sönke ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Can Haskell outperform C++?
Chris, Thanks you for indulging me. I will eventually get use to the idea that there is a wiki page for almost every topic :-) Gregg On 5/12/2012 1:02 AM, Chris Wong wrote: On Sat, May 12, 2012 at 12:41 AM, Gregg Lebovitzglebov...@gmail.com wrote: I would find it useful to pull all this information together into a single document that discusses all the performance issues in one place and shares the real life experience is dealing with each issue. I see this as a best practice paper rather than a research document. Does such a document exist? If not, I am willing try and start one. http://www.haskell.org/haskellwiki/Performance ;) ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Linear Diaphantine equation solver bug
Recall I reported a surprising result from an implementation of a Linear Diophantine equation solver based on an algorithm by Contejean and Devie. When using it to solve an inhomogeneous equation, a version of my code generated a non-minimal solution. It turns out that when solving the inhomogeneous problem, one must filter out solutions to the homogeneous problem. This was not clear from the description of the algorithm I used, but a careful analysis of the section on inhomogeneous equations in the Contejean and Devie papers makes clear the correct fix. In other news, Levent Erkok sent news about his SVB package. John Levent: In case this is of any interest to you; the latest release of SBV now has a diophantine solver example: http://hackage.haskell.org/packages/archive/sbv/2.0/doc/html/Data-SBV-Examples-Existentials-Diophantine.html It's generalized to a system of equations instead of just one, and uses the underlying SMT solver to find the minimal solutions. I see it as a nice complement to your work, where one can check the results of your solver against the SBV produced one for extra assurance if needed. On Wed, Apr 25, 2012 at 7:09 PM, John D. Ramsdell ramsde...@gmail.com wrote: I uploaded a new version of the ACU unifier in package cmu. It includes a Linear Diophantine equation solver that now handles inhomogeneous equations. What's interesting is the algorithm is based on a paper by Contejean and Devie. That paper includes a proof of correctness of their algorithm. Yet I found an example in which following the steps in the order they gave, the algorithm produces extra answers. I had to switch the order of the steps in the algorithm to get the right answer. I wonder if anyone else knows of this issue. The example is given in the module documentation on Hackage. John http://hackage.haskell.org/packages/archive/cmu/1.7/doc/html/Algebra-CommutativeMonoid-LinDiophEq.html ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] ANN: Patterns 0.0.3
Hi! I am pleased to announce patterns-0.0.3. On Hackage: http://hackage.haskell.org/package/patterns The Patterns package implements communication patterns that are often used in distributed, message-oriented applications. The package implements a set of basic patterns and a device to connect basic patterns through routers, brokers, load balancers, etc. The package is based on the zeromq library, but abstracts from the notion of sockets, on which zeromq is built. Cheers Tobias ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] darcs patch dependencies in dot format
Truly amazing! I wonder it would fare with larger repositories. =) Cheers, -- Felipe – enviado do meu Galaxy Tab. Em 12/05/2012 09:52, Sönke Hahn sh...@cs.tu-berlin.de escreveu: Hi all! Yesterday I wrote a little tool to output the dependencies of darcs patches in dot format. The hardest part was to wrap my head around the darcs API and find a way to let it compute the patch dependencies. I don't know, if I got it right, but it looks correct at first glance. Here is the code: https://patch-tag.com/r/shahn/darcs2dot To use it just execute the program in a darcs repo and it will output a dot graph to stdout. The graph does not contain all dependencies, but the transitive reduction. The patch names are truncated at 15 characters. And here is an example graph: http://open-projects.net/~shahn/patchDeps.pdf These are the patch dependencies of one of my darcs repos (https://patch-tag.com/r/shahn/hate). Some observations I made: * There are two completely separate subgraphs. One subgraph seems to be for the testsuite, the other for the actual code. This means, the two don't depend on each other and could easily be put in two distinct repos. * There is a re-implementation patch with a lot of incoming and outgoing edges. (Which makes sense.) * There are some sequences of patches (e.g. these four allow ... patches in the upper left corner) that seem to contain related patches. * darcs's patch system is awesome! Any comments or suggestions? Cheers, Sönke ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Can Haskell outperform C++?
Gregg Lebovitz glebov...@gmail.com wrote: Ryan writes: With a few years of Haskell experience in my backpack I know how to utilize laziness to get amazing performance for code that most people would feel must be written with destructively updating loop. That was me actually. Just a minor side note that some people may consider a nitpick, but I'd like to keep being the author of my own statements. Thanks. Greets, Ertugrul -- nightmare = unsafePerformIO (getWrongWife = sex) http://ertes.de/ signature.asc Description: PGP signature ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe