Comments on: Lazy functional yak shaving in Haskell http://www.serpentine.com/blog/2009/09/12/lazy-functional-yak-shaving-in-haskell/ Bryan O'Sullivan's blog Wed, 08 Feb 2012 06:41:38 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Tom D http://www.serpentine.com/blog/2009/09/12/lazy-functional-yak-shaving-in-haskell/comment-page-1/#comment-246181 Tom D Fri, 18 Sep 2009 16:44:07 +0000 http://www.serpentine.com/blog/?p=424#comment-246181 Thanks for the library Brian, once more very interesting code for us all to study and use. I use R for statistical work all the time, it is robust, efficient and has packages for practically everything. At some point I'm sure (/I hope) someone will write the necessary binding layer so that Haskell can call into R via the direct C api or else through Rserve. However, I think we'll also benefit from having a decent native Haskell implementation of standard statistical functionality; for one thing it means we won't need to depend on an external library like R, and the functions will be easier for Haskell coders to use than requiring them to learn R, and the unique features of Haskell will provide many benefits to the implementation So I'm rather confused as to what the "correct" configuration is, and would like to know what you think of this? It seems that the way that applications and libraries are structured, the oss community is destined to massive duplication of effort, again and again. Every language gets its own implementation of statistics functionality, some better than others, often coded by people with limited stats knowledge doing their best to add to their favourite language, even though we already have this functionality, written by statisticians. Think of the thousands of hours spent debugging and refining all the various implementations of essentially the same functionality... of course this isn't unique to statistics libraries, BLASs are another pertinent example and there are dozens more. Sorry for the long comment, and looking forward to the benchmarking lib. Thanks for the library Brian, once more very interesting code for us all to study and use.

I use R for statistical work all the time, it is robust, efficient and has packages for practically everything. At some point I’m sure (/I hope) someone will write the necessary binding layer so that Haskell can call into R via the direct C api or else through Rserve. However, I think we’ll also benefit from having a decent native Haskell implementation of standard statistical functionality; for one thing it means we won’t need to depend on an external library like R, and the functions will be easier for Haskell coders to use than requiring them to learn R, and the unique features of Haskell will provide many benefits to the implementation

So I’m rather confused as to what the “correct” configuration is, and would like to know what you think of this? It seems that the way that applications and libraries are structured, the oss community is destined to massive duplication of effort, again and again. Every language gets its own implementation of statistics functionality, some better than others, often coded by people with limited stats knowledge doing their best to add to their favourite language, even though we already have this functionality, written by statisticians. Think of the thousands of hours spent debugging and refining all the various implementations of essentially the same functionality… of course this isn’t unique to statistics libraries, BLASs are another pertinent example and there are dozens more.

Sorry for the long comment, and looking forward to the benchmarking lib.

]]> By: Johan Tibell http://www.serpentine.com/blog/2009/09/12/lazy-functional-yak-shaving-in-haskell/comment-page-1/#comment-246014 Johan Tibell Tue, 15 Sep 2009 08:44:41 +0000 http://www.serpentine.com/blog/?p=424#comment-246014 I wonder if timsort can be made to work in the Unicode case. http://en.wikipedia.org/wiki/Timsort I wonder if timsort can be made to work in the Unicode case.

http://en.wikipedia.org/wiki/Timsort

]]> By: mightybyte http://www.serpentine.com/blog/2009/09/12/lazy-functional-yak-shaving-in-haskell/comment-page-1/#comment-245960 mightybyte Mon, 14 Sep 2009 13:49:48 +0000 http://www.serpentine.com/blog/?p=424#comment-245960 Doesn't the Boyer-Moore string searching algorithm have a complexity of O(n/m) in the average case and O(n+m) in the worst case? Doesn’t the Boyer-Moore string searching algorithm have a complexity of O(n/m) in the average case and O(n+m) in the worst case?

]]> By: solrize http://www.serpentine.com/blog/2009/09/12/lazy-functional-yak-shaving-in-haskell/comment-page-1/#comment-245933 solrize Sun, 13 Sep 2009 23:09:29 +0000 http://www.serpentine.com/blog/?p=424#comment-245933 By the way your stats library looks very interesting. I'm going to try understanding it, despite being somewhat stats-clueless (but trying to change that). Some tutorial docs (or anyway some pointers to where to find more info) would probably be helpful, though I guess I can start with Wikipedia. By the way your stats library looks very interesting. I’m going to try understanding it, despite being somewhat stats-clueless (but trying to change that). Some tutorial docs (or anyway some pointers to where to find more info) would probably be helpful, though I guess I can start with Wikipedia.

]]> By: solrize http://www.serpentine.com/blog/2009/09/12/lazy-functional-yak-shaving-in-haskell/comment-page-1/#comment-245932 solrize Sun, 13 Sep 2009 23:06:16 +0000 http://www.serpentine.com/blog/?p=424#comment-245932 The floating point article at sun.com appears to be a reprint of a Computing Surveys article by David Goldberg: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.22.6768 For some reason Sun cites Computing Surveys but doesn't name the original author of the article. They also say it was reprinted by permission but also have an enormous blob of legalbol claiming it is copyright by Sun! The floating point article at sun.com appears to be a reprint of a Computing Surveys article by David Goldberg:

http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.22.6768

For some reason Sun cites Computing Surveys but doesn’t name the original author of the article. They also say it was reprinted by permission but also have an enormous blob of legalbol claiming it is copyright by Sun!

]]>