- Haskell-Cafe - Haskell.org

Re[2]: [Haskell-cafe] request for help
by Bulat Ziganshin 04 Feb '06

04 Feb '06

Hello Marco, Saturday, February 04, 2006, 4:39:23 PM, you wrote: MAFdA> Hello. MAFdA> English is not my native language (I'm Portuguese). MAFdA> The haskell-cafe mailing list is usually very active and subscribers MAFdA> do help each other, so I believe you will not have any trouble in MAFdA> finding "real" Englishmen who will read and correct the document ;-) thank you. J.P. Bernardy already corrected a part, so i think that Englishmens will help me MAFdA> However, if by any chance you don't, please let me know and I will do it MAFdA> (just tell me how, because the wiki is password-protected). you just need to register itself in this Wiki system. look at the "register" text on the page, i don't remember exact. or may be "login", where you will have chance to register -- Best regards, Bulat mailto:bulatz@HotPOP.com

1 0

Re: Haskell code for this example of flow control
by Dominic Steinitz 04 Feb '06

04 Feb '06

Here are some even older discussions on the subject. I don't know if anyone ever put them into a library or on the wiki. Dominic. http://haskell.org/pipermail/haskell-cafe/2005-May/009784.html http://www.haskell.org//pipermail/libraries/2005-February/003143.html

1 0

request for help
by Bulat Ziganshin 04 Feb '06

04 Feb '06

Hello haskell-cafe, i'm wrote new general i/o library called Streams. it's so great that i hope it will eventually replace using Handles. i plan to present it in Haskell list on Monday. in order to do it i wrote overview of library facilities. the problem is that i'm not native english speaker and i feel that my text is awkward, in best case. when i wrote last time docs about Arrays, Cale Gubbard fixed lots of my misspellings on wiki page :) this time i try to go different way - i published my version of report at the http://haskell.org/haskellwiki/Library/Streams page and ask your help in correcting it. it will allow to include in presentation the proper English text that is much more pleasing for readers than my scribbles :) so, please correct text at this wiki according to English rules. even if you correct paragraph or two, that will forward the work further -- Best regards, Bulat mailto:bulatz@HotPOP.com

1 0

Pythonic monads
by Graham Klyne 03 Feb '06

03 Feb '06

Constructing some code today in Python, using some functional-style coding idioms, I found myself wondering if there would be any real benefit to using a monad-based implementation (i.e. other than to demonstrate that it can be done). The application that sparked this line of thought was a simple filter to trim comments and whitespace out of an XML document. I ended up with a simple state-machine driven filter, thus: [[ # Strip comments from XML strings def stripXmlComments(x): # State table for stripping XML comments stStripXmlComments = \ { 0: match1('<',(1,stateFilterHold),(0,stateFilterPass)) , 1: match1('!',(2,stateFilterHold),(0,stateFilterPass)) , 2: match1('-',(3,stateFilterHold),(0,stateFilterPass)) , 3: match1('-',(4,stateFilterDrop),(0,stateFilterPass)) , 4: match1('-',(5,stateFilterDrop),(4,stateFilterDrop)) , 5: match1('-',(6,stateFilterDrop),(4,stateFilterDrop)) , 6: match1('>',(0,stateFilterDrop),(4,stateFilterDrop)) } return stateFilter(stStripXmlComments,x) # Simple state machine driven filter # # The state table is a dictionary indexed by state values, where the # initial state is 0, and each entry is a function that accepts a next # symbol and returns a pair of (next state, action), where action is # one of 'stateFilterPass', 'stateFilterDrop', 'stateFilterHold'. # stateFilterHold means that the disposition will be determined later. # # The result is an iterator that returns elements from the filtered # subsequence of the supplied sequence. # def stateFilter(stable,seq): queue = [] state = 0 for symbol in seq: (state,action) = stable[state](symbol) (queue,emit) = action(queue,symbol) for e in emit: yield e return def stateFilterPass(q,n): return ([],q+[n]) def stateFilterDrop(q,n): return ([],[]) def stateFilterHold(q,n): return (q+[n],[]) # State transition function to match the specified symbol and return # 'eqval' if matched, otherwise 'neval' def match1(sym,eqval,neval): def m(sym,eqval,neval,next): if next==sym: return eqval return neval return curry(m,sym,eqval,neval) def curry(func, *args): """ Curry multiple arguments: See: http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/229472 """ def curried(*args2): args2 = args + args2 return func(*args2) return curried ]] and a test case: [[ def testFilter02(self): fullstr = "<!-abc->  <!- -ghi-->" trimstr = list(stripXmlComments(fullstr)) expstr = list("<!-abc-> <!- -ghi-->") assert trimstr==expstr, \ "stripSpaces, expected:\n"+expstr+"\nFound:\n"+trimstr ]] In thinking about this implementation, it seemed to me that this employed patterns characteristic of a monadic type: each entry in the state table (in this case, an instance of match1, a curried function) is like a step in a monadic computation, updating the monadic value and also returning some value. What I can't quite visualize is if the code in Python would actually look any better if it were implemented with a monadic type, as one might readily choose for a Haskell implementation. Or would there be no real benefit? I have noticed that, while I like to use functional idioms in some of my Python code, and the Python language is easily able to support these (even some lazy evaluation, courtesy of generators), that the code doesn't always look as clean as its Haskell equivalent. In Haskell, composition and currying are fundamental patterns and are directly supported by the syntax. In Python, one has to work harder to achieve these (e.g. the "curry" function above seems rather convoluted to me, for such a fundamental notion). Thoughts? Comments? #g -- Graham Klyne For email: http://www.ninebynine.org/#Contact

2 1

Haskell code for this example of flow control
by Maurício 03 Feb '06

03 Feb '06

Hi, I would like to know what options I have in Haskell to do something similar to this C++ code: double a = 1000; while (a>1) a/=2; I'm able to do that with lists, but I would like to know how to do that with monads and variables with state. Thanks, Maurício

7 9

HUnit
by Matt Roberts 03 Feb '06

03 Feb '06

I am in love with HUnit and QuickCheck. However, I am missing one important functionality I want to call a command line program that will search all the .hs files I have in a particular directory, pull out all HUnit test (based on what they are called, or perhaps their type signature) and run them all. I have done a bit of a search and can't find anything like this, am I going to have to write it myself? Matt

3 2

Known Unknowns
by Chris Kuklewicz 03 Feb '06

03 Feb '06

There is a new combined benchmark, "partial sums" that subsumes several earlier benchmarks and runs 9 different numerical calculations: http://haskell.org/hawiki/PartialSumsEntry I took the Clean entry and, since there were so many recursive computations, I made compute and altCompute higher order functions. No speed optimization work has been done yet, but it seems really fast. A new http://haskell.org/hawiki/KnucleotideEntry has been accepted, replacing the one with the memory leak. The speed is also better, but still poor, due to Data.Hashtable. GHC 6.6.x should eventually help. Also on the topic of poor speed: http://haskell.org/hawiki/NbodyEntry At the bottom of the page is Einar's really cool existing entry (open mutable records). At the top of the page is best effort to replace it, but it runs at almost the same speed. Which is very much slower (10x) than other entries, such as OCaml. It is odd that small scale numerics, such as the partial-sums functions, can be made fast but not medium scale numerics like the n-body integration. There are 5 fixed mass parameters and 30 variables (5*3 positions + 5*3 velocities). Addition, Subtraction, Multiplication, Division. And we can't get it going nearly as fast as it ought to run. Note: The garbage collector is not a factor (%GC time 0.9%). Like the Hashtable glitch, if this were part of real program then a small bit of c-code could be attached via FFI. Are there little known Haskell idioms for doing math? Do we have to write a recursive function with 30 separate parameters? Do we have to peek and poke unboxed doubles? Is there an undocumented -funleash-fortran parameter? Will this be fixable in Haskell' ("Haskell-Prime")?

7 18

Number 1, at least for now
by dons＠cse.unsw.edu.au 02 Feb '06

02 Feb '06

Haskell is now ranked number 1 on the Great Language Shootout! http://shootout.alioth.debian.org/gp4/benchmark.php?test=all&lang=all Hooray :) -- Don

8 12

Re: [Haskell-cafe] EclipseFP (Haskell IDE) 0.9.1 released
by dagit＠eecs.oregonstate.edu 02 Feb '06

02 Feb '06

Quoting Thiago Arrais <thiago.arrais(a)gmail.com>: > 2006/2/1, Graham Klyne <GK(a)ninebynine.org>: > > [I should say that I haven't yet actually tried this software, though I'd > like > > to do so when I get some time.] > > I really hope you find the time to do so. We would be glad to be > hearing from you again. I haven't found time (yet) to look at your new version, but I tried out the stable version from about a week ago and played with it a bit. I might as well tell you what I found then and why I'm unable to use EclipseFP at this time. > At some point of the wave, the integrated environments appear. They > come, as the name says, to integrate the previous tools in an easy to > use and productive environment. Not that the previous tools were hard > to use or counter-productive, it is just that by integrating them much > time is saved from the details and the programmer can spend more time > on the creative and useful things only he can do. I think it's important to keep this aspect, the ramping up and the integration. Particularly, I'm against tools that want to do things their way. I'd much rather have the tool automate the tools I already have instead of creating a new mechanism. For example, I'd rather have the IDE work with my makefiles than ignore them and use some builtin make system. I need to be able to separate from the IDE whenever it becomes cumbersome or when I want to share the code. > > So is there a compelling feature in this Eclipse plugin that isn't easily > > achieved using simpler tools? > > When we write an Eclipse plugin, we get a lot of things for free. Just > to cite one, there is already CVS integration support within every > Eclipse installation, which includes a very neat diff viewer. Adding > support for other version control systems isn't very hard. There are > plugins, for example, for Darcs and Subversion too. I was trying the Darcs plugin in conjunction with EclipseFP as my first attempt at using eclispe. I found that waiting for eclipse to "autobuild" everything in _darcs was a waste of time and just plain silly. Hopefully, there is a way to avoid this excessive building and still use autobuild. I couldn't find it. > Another very nice feature of the Eclipse platform is the refactoring > support. Wouldn't you like to refactor your Haskell code as easily as > selecting some context-menu item? Not to mention code assistance and > 'go to declaration support' (click a module/function/whatever > reference and open its corresponding declaration). This has saved me a > lot of time when browsing code. This is the reason I sought out EclipseFP. I want to automate as much of the refactoring as I can while making it safe. But now let me turn to the reasons I'm not using it anymore. Some of the reasons have to do with eclipse itself (I think), some are related to the very nice plugin you have created, and some are related to the darcs plugin. Some of my problems may even be fixed in the announced version, but I haven't had a chance to check. I wanted to use EclipseFP to help with darcs development. The first problem I encountered was that darcs has a makefile which is growing in complexity. Some files are compiled with this flag, others with that flag. I couldn't figure out how to import the make rules. I did find a way to set the compilation options, but they were per project instead of per file as near as I could tell. I would a makefile import to be automated and to watch the makefile for changes. One direction would be fine for now, but ultimately I would want it to let me change build settings in Eclipse and have it mirrored in the makefile. Does eclipse have a makefile plugin? I also had a problem with autobuild building too much as noted above. I also couldn't figure out a way to ask eclipse to delegate to make for building. The next problem I encountered was the parser was not forgiving enough to parse the darcs source code, as a result everything was colored as a comment. Every file that contained "#include", "#define" or similar threw a lexical error. The code is meant to be run through ghc with the -cpp flag turned on. Is it possible to use ghc-api for parsing the code? I also noticed that when you click on an error so that eclipse will take you to the place of the error, it often took me to lines that were perfectly fine, in fact, I think it was often the wrong file. The first time I created a "Hello, World" project in win32 eclipseFP gave some odd errors that I didn't write down. I haven't seen that error since so I can't make a bug report, but it was certainly confusing. It was to the effect that ghc had crashed while generating the project. After I got that error, I stop using eclipse in windows and switched to OSX. I had some other problems, but don't think they were related enough to list them here. In the end, it was a horrible first experience with eclipse and I'm not looking forward to ever using eclipse again. On the other hand, I want some sort of IDE for Haskell to help automate certain things, like refactoring. I'm a long time fan of emacs, but the more I use emacs the more I see that it may never handle certain things well. For example, using regexp to do syntax highlighting is quick and easy it's also the source of many of the problems with using emacs for coding in Haskell. The indentation is off, block uncomment is not the inverse of block comment, if you use a string like "{-" it throws off the highlighting until a matching -} is found and so on. My advice would be to download the darcs source code and try to use darcs with EclipseFP. You'll see first hand some things that could use some fix'n :) And I think I could stand to use eclipse to test out your changes if you need a tester, but I doubt I'd have much more time to spare than testing. > Of course, you need to write the code to tell the platform about your > language. That's what the EclipseFP team is trying to do. Yes, and I'm very happy to see someone doing this. I wish you the best of luck and when the problems I encountered are fixed I will try eclipse again. Hopefully my rantings will be helpful to you and not make you lose hope :) I really want see you succeed. Thanks, Jason

2 1

RE: [Haskell-cafe] Number 1, at least for now
by Simon Peyton-Jones 02 Feb '06

02 Feb '06

| > though, I think this is a great oprotunity to improve ghc's optimizer. | | Huh, that would be the best thing with the whole shootout endeavour.. Yes indeed. One thing that would be really helpful, as a first step, would be to identify a bunch of concrete examples that GHC "should" have optimised, or perhaps "could" have optimised, and attach them to the GHC wiki somewhere. One place would be here http://haskell.org/haskellwiki/Performance or perhaps somewhere on the developers wiki, or as a task. http://hackage.haskell.org/trac/ghc I'm not promising that anything will happen fast (unless someone else does it!) but it's always much more likely to happen if there's a specific place that advertises low-hanging performance fruit. This is closely related to the "advice to programmers" that Don circulated yesterday, in which he explained how he optimised one particular case. But not quite the same. Simon

1 0