Re: [Haskell-cafe] Cleaning up stable names?
but without that function, stable names are not that useful I guess? they would cause a space leak? On Mon, Aug 31, 2009 at 10:59 PM, Job Vranish jvran...@gmail.com wrote: I also would like a isStableNameTargetAlive function. Though if you had such a function then you probably _could_ make a deRefStableName function, which, since there isn't one, probably means that such a function would be hard to make. - Job On Sun, Aug 30, 2009 at 4:54 PM, Peter Verswyvelen bugf...@gmail.comwrote: From the documentation, I don't think I grasp how stable names work. From the docs: There is no deRefStableName operation. You can't get back from a stable name to the original Haskell object. The reason for this is that the existence of a stable name for an object does not guarantee the existence of the object itself; it can still be garbage collected. From this I can conclude that stable names behave a bit like weak pointers. However, suppose I have a hash table of these stable names. How can I remove the redundant stable names from the table? I mean removing stable names that refer to an object that is garbage collected? I don't see any function for checking that (e.g. isStableNameTargetAlive or something) Thanks, Peter ___ 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
[Haskell-cafe] Re: Snow Leopard Breaks GHC
It seems, bootstrapping cabal went wrong. Does http://hackage.haskell.org/platform/2009.2.0.2/haskell-platform-2009.2.0.2-i386.dmg work? Once cabal works, options --ld-option=-m32 (and also --gcc-option=-m32) may be used. These options may also be passed to ./Setup configure C. Tom Tobin wrote: Hmm ... running Snow Leopard here, I added these arguments to my /usr/bin/ghc (and my /usr/bin/ghci, and /usr/bin/runhaskell, for good measure), and I end up getting the following error on cabal update after installing cabal-install: * cabal: user error (Codec.Compression.Zlib: incompatible version) * Earlier in the cabal-install bootstrap process, I get the following line: * ld: warning: in /Library/Frameworks/GHC.framework/Versions/610/usr/lib/ghc-6.10.4/libgmp.a, file is not of required architecture * I'm guessing something's still not being set to 32-bit? ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Takusen - is anyone currently using it on Win32 - ODBC?
2009/9/1 Günther Schmidt gue.schm...@web.de: One of the files for instance has 298 k lines. A glance over sushi revealed that parsec is involved so I can only presume it is read into memory all at once. That would certainly be a problem. There is nothing inherent about parsec that would cause all input to be read into memory, TxtSushi uses parsec for parsing SQL, not for CSV, by the way. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Question about Lazy.IO
Hi, I've been wondering about Lazy IO for a while. Suppose we have a program like main = interact (unlines . somefunction . lines) then somefunction is a pure function. With the semantic interpretation of: given a input list return an output list. However I, the user, can change my input depending on the output of this function. Now in simple cases like this, this will not be a problem. But suppose you are reading and writing to a file. Now the result of pure functions become dependent on the order of execution, breaking (I think) referential transparency. Am I wrong here or how could you prove that Lazy IO is consistent nonetheless? Greetings, Gerben -- View this message in context: http://www.nabble.com/Question-about-Lazy.IO-tp25236848p25236848.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Question about Lazy.IO
On Tue, Sep 1, 2009 at 3:05 AM, staafmeisterg.c.stave...@uu.nl wrote:
Hi,
I've been wondering about Lazy IO for a while. Suppose we have a program
like
main = interact (unlines . somefunction . lines)
then somefunction is a pure function. With the semantic interpretation of:
given a input list
return an output list. However I, the user, can change my input depending on
the output of this
function. Now in simple cases like this, this will not be a problem. But
suppose you are reading
and writing to a file. Now the result of pure functions become dependent on
the order of execution,
breaking (I think) referential transparency. Am I wrong here or how could
you prove that Lazy IO
is consistent nonetheless?
Ignoring timeliness in responses (which our theory doesn't talk
about), you are allowed to change your input based on its output in
exactly the same way as any other function could.
The reason this is okay is in the realm of domain theory. There is a
good introductory tutorial in the Haskell wikibook:
http://en.wikibooks.org/wiki/Haskell/Denotational_semantics
Here is an intuition. Let's forget you are a user, and just call you
a function user.
user :: String - String
program :: String - String
The input and output of a program is related to these functions like so:
let input = user output
output = program input
in (input, output)
Now, user could certainly be a function like:
user ('a':rest) = 'x':rest
user ('b':rest) = 'y':rest
Which makes a decision about what to type before seeing all the
output. Program could make use of this:
program inp = 'a':case inp of { 'x':_ - hello; 'y':_ - world }
This mutual recursion would not be well-defined if all of the output
must be seen before the input can be known, because the output depends
on the input and the input depends on the output. However, this is a
perfectly valid Haskell program, without considering I/O, and will
compute (input, output) to be (xhello, ahello).
I have now tried twice to explain why this is possible more deeply in
terms of domain theory, but it has bloated the message to three times
this size. If you would like me to include my response, I'd be more
than happy to, but it does get rather technical.
Luke
___
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Question about Lazy.IO
Dear Luke, Thanks for your reply. I am interested in a more detailed explanation. If it's to long for this forum you can also mail it to me. In response to your answer. In your program the user is also a pure function. But I, as a user, am not pure so I can do side effects. More specific suppose I program something like this ss - hGetContent handle (lazy IO) dosomethingweird handle ss (a function that alters the file at will) Now as a programmer I should be positive that ss will be the file as it was before dosomethingweird modifies it. But this can never be guaranteed if ss is lazy. And now things start to depend on the order of evaluation breaking referential transparency, or do I overlook something. For concreteness suppose dosomethingweird appends ss to the end of the file. If ss is strict then it just copies the whole file, if ss is lazy it copies the whole file an infinite number of times. Gerben -- View this message in context: http://www.nabble.com/Question-about-Lazy.IO-tp25236848p25238044.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Problems installing documentation and sources using Cabal on Mac OSX
I noticed today on the train (no internet connection!) that the documentation for packages I had installed with cabal was missing, and that there was no source in .cabal/packages for them. Now I'm in the offiice, I tried --reinstall --enable-documentation. There was a warning about haddock and ghc version mismatch (so much for the haskell platform install), so I did a --reinstall of haddock. Now the message I see is: cabal: can't find source for module Paths_happstack_server (happstack_server is the first package I'm trying to reinstall to get docs). So I think the documentation failure is because haddock can't find the source code (nor can I), but I don't know why there is no source installed in .cabal/packages - what should I do to further track this problem down? -- Colin Adams Preston, Lancashire, ENGLAND ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: Re[2]: [Haskell-cafe] GUI library
Deallocation is automatic (just like C++ Qt) C++ Qt has excellent unicode support. I haven't explicit tried it in qtHaskell, but as far as I know it should work just fine. Unfortunatly it does not currently build just with cabal, however the build scripts are very clean (no configure) and the haskell half is cabalized. To build on windows you only need the haskell platform and the Qt SDK (and then you either have to put the ghc path into Qts qtEnv.bat (what I did), or you have to put the Qt paths into the system path) and then you run the qtHaskells build.bat and you're done. To build on linux you need the haskell platform and Qt4 dev libraries (they are probably in your distros package manager) and then you run the qtHaskells build script, and your done. Umm, with one gotcha, on my system I had to replace two instances of 'gmake' in the build script with 'make'. You might not have to do that on your system. Qt uses the native APIs to draw controls, so your apps look appropriate on different platforms. - Job On Tue, Sep 1, 2009 at 12:43 AM, Bulat Ziganshin bulat.zigans...@gmail.comwrote: Hello Job, Tuesday, September 1, 2009, 1:16:38 AM, you wrote: I recommend qtHaskell. how it's in areas of - memory deallocation when it's no more need - is it automatic or manual? - unicode support - compatibiliy with latest ghc versions - does it build by Cabal or we need to wait while gurus release installers for Windows? - does it use native GUI controls on Windows or draws them itself? i'm asking because those questions arise when using other libs -- Best regards, Bulatmailto:bulat.zigans...@gmail.com ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] GUI library
If you're already used to C++ Qt and PyQt, qtHaskell should be relatively straightforward (though probably with a few gotchas). You could download qtHaskell and look at the examples to get an idea for the feel of it. If you're already used to Qt they should look familiar. Subclassing _is_ done a little weird: First you declare a dummy datatype: data MyQPushButton = MyQPushButton And then you use the qSubClass function: myQPushButton :: String - IO (QPushButton MyQPushButton) myQPushButton s = qSubClass $ qPushButton1 s The type signature here is necessary. It's the only thing that forces the new (QPushButton a) to be a QPushButton MyQPushButton. Then you can use your subclassed buttons like so: main :: IO () main = do app - qApplication dialog - qDialog button1 - myQPushButton Click for Stuff qObject_connectSlot1 button1 clicked() button1 click() $ on_pbutton_clicked dialog mainLayout - qVBoxLayout qLayout_addWidget mainLayout button1 qWidget_setLayout dialog mainLayout qWidget_setWindowTitle dialog Stuff Test ok - qDialog_exec dialog return() on_pbutton_clicked :: QDialog () - QPushButton MyQPushButton - IO () on_pbutton_clicked _dlg _this = do mb - qMessageBox1 _dlg qMessageBox_setText mb $ Stuff qWidget_show mb return () Hmmm, though it doesn't look like you can currently overload methods of your parent class, usually you don't need too though as you can just tie into a signal (so you still can do things like paint). Though I'm kinda suprised I didn't notice this before, I'll have to email the guy and see if this is actually an issue. Also, qtHaskell doesn't fully support all the Qt widgets (though it covers most of them) but there is a new version coming out soon that should be more complete. Overall qtHaskell has served my perposes well. There is also a good listing of GUI toolkits for Haskell at http://www.haskell.org/haskellwiki/Applications_and_libraries/GUI_libraries - Job On Tue, Sep 1, 2009 at 1:49 AM, Michael Mossey m...@alumni.caltech.eduwrote: Thanks for the info. Interesting. I'm already familiar with C++ Qt and also PyQt. I am also a fan of Qt. However, as a beginner to Haskell, I want to be sure I don't get myself into trouble. I hope that qtHaskell is straightforward. Qt depends on deriving user classes. How is this handled in qtHaskell? Thanks, -Mike Job Vranish wrote: I recommend qtHaskell. I am a big fan of Qt in general. It has good documentation and extensive examples, is very well designed, and has a good license. I'd even say the C++ version is good choice for beginners (certainly easier to understand/use than say GTK). The qtHaskell bindings are also pretty good. The documentation and examples are not as extensive, but you can usually use the C++ documentation to fill in the gaps. Being already familiar with C++ Qt, using qtHaskell was a snap. However, if you're unfamiliar with both Qt and Haskell it will probably be confusing at first. Though I'd bet money the GTK bindings aren't any better in that regard. I'd still say you'd be more productive with qtHaskell in the long run. - Job On Sat, Aug 29, 2009 at 11:03 AM, Michael Mossey m...@alumni.caltech.edumailto: m...@alumni.caltech.edu wrote: I want to choose a GUI library for my project. Some background: I'm a beginner to functional programming and have been working through Haskell books for a few months now. I'm not just learning Haskell for s**ts and giggles; my purpose is to write music-composition-related code; in particular, I want to write a graphical musical score editor. (Why write my own editor, you may ask? Because I want to fully integrate it with computer-assisted-composition algorithms that I plan to write, also in Haskell.) I decided to use Haskell for its great features as a functional programming language. Regarding a choice of GUI library, I want these factors: - it needs to provide at a minimum a drawing surface, a place I can draw lines and insert characters, in addition to all the standard widgets and layout capabilities we have to come to expect from a GUI library. - This is a Windows application. - it needs to be non-confusing for an intermediate-beginner Haskeller. Hopefully good documentation and examples will exist on the web. - It might be nice to have advanced graphics capability such as Qt provides, things like antialiasied shapes, and a canvas with efficient refresh (refereshes only the area that was exposed, and if your canvas items are only primitives, it can do refreshes from within C++ (no need to touch your Haskell code at all). However I'm wondering if qtHaskell fits my criteria well-documented and lots of examples aimed at beginners. Thanks, Mike ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org
Re: [Haskell-cafe] Cleaning up stable names?
Well usually, when I've used stable names, I've just used them to check if things are the same, and then thrown them away. So no chance for a space leak. It's usually unsafe to keep stable names around for very long as they can lose their ability to tell if two things are the same (if this surprises you, you should carefully reread http://haskell.org/ghc/docs/latest/html/libraries/base/System-Mem-StableName.html#v%3AmakeStableName). Out of curiosity, how are you planning on using them? - Job On Tue, Sep 1, 2009 at 3:40 AM, Peter Verswyvelen bugf...@gmail.com wrote: but without that function, stable names are not that useful I guess? they would cause a space leak? On Mon, Aug 31, 2009 at 10:59 PM, Job Vranish jvran...@gmail.com wrote: I also would like a isStableNameTargetAlive function. Though if you had such a function then you probably _could_ make a deRefStableName function, which, since there isn't one, probably means that such a function would be hard to make. - Job On Sun, Aug 30, 2009 at 4:54 PM, Peter Verswyvelen bugf...@gmail.comwrote: From the documentation, I don't think I grasp how stable names work. From the docs: There is no deRefStableName operation. You can't get back from a stable name to the original Haskell object. The reason for this is that the existence of a stable name for an object does not guarantee the existence of the object itself; it can still be garbage collected. From this I can conclude that stable names behave a bit like weak pointers. However, suppose I have a hash table of these stable names. How can I remove the redundant stable names from the table? I mean removing stable names that refer to an object that is garbage collected? I don't see any function for checking that (e.g. isStableNameTargetAlive or something) Thanks, Peter ___ 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] Cleaning up stable names?
I was planning to use them for caching OpenGL display lists and render targets. These are generated from a pure scene graph description, and if the same description is handed over to the render engine, it would just reuse the previously cached OpenGL object. I can of course just embedding and IORef inside the pure structures, that would also work I guess On Tue, Sep 1, 2009 at 6:09 PM, Job Vranishjvran...@gmail.com wrote: Well usually, when I've used stable names, I've just used them to check if things are the same, and then thrown them away. So no chance for a space leak. It's usually unsafe to keep stable names around for very long as they can lose their ability to tell if two things are the same (if this surprises you, you should carefully reread http://haskell.org/ghc/docs/latest/html/libraries/base/System-Mem-StableName.html#v%3AmakeStableName ). Out of curiosity, how are you planning on using them? - Job On Tue, Sep 1, 2009 at 3:40 AM, Peter Verswyvelen bugf...@gmail.com wrote: but without that function, stable names are not that useful I guess? they would cause a space leak? On Mon, Aug 31, 2009 at 10:59 PM, Job Vranish jvran...@gmail.com wrote: I also would like a isStableNameTargetAlive function. Though if you had such a function then you probably _could_ make a deRefStableName function, which, since there isn't one, probably means that such a function would be hard to make. - Job On Sun, Aug 30, 2009 at 4:54 PM, Peter Verswyvelen bugf...@gmail.com wrote: From the documentation, I don't think I grasp how stable names work. From the docs: There is no deRefStableName operation. You can't get back from a stable name to the original Haskell object. The reason for this is that the existence of a stable name for an object does not guarantee the existence of the object itself; it can still be garbage collected. From this I can conclude that stable names behave a bit like weak pointers. However, suppose I have a hash table of these stable names. How can I remove the redundant stable names from the table? I mean removing stable names that refer to an object that is garbage collected? I don't see any function for checking that (e.g. isStableNameTargetAlive or something) Thanks, Peter ___ 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
[Haskell-cafe] Can't get my head round monad transformers
I'm trying to add a state monad onto the IO monad for use in a happstack application. I thought this should involve using StateT and Happstack.Server.SimpleHTTP.simpleHTTP', but I can't figure out how to plumb it all together. Any tips will be welcome. -- Colin Adams Preston, Lancashire, ENGLAND ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Slow IO?
2009/08/31 Don Stewart d...@galois.com: If you can abstract out a common function for lexing ints out of bytestrings, we could add it to the bytestring-lexing package. All the really performant implementations operate on strings with multiple ints in them; I suspect this reduces memory traffic -- and indeed, Eugene's code using my libs allocates about twice as much memory as Don's code. I've tried a few different things with strictness annotations to no avail. I'm having some trouble understanding the meaning of entries in the profiler's output. I have a file with 5 million random integers in it, totalling 26210408 bytes (21210408 bytes of which are not newlines). The relevant part is here: COST CENTRE MODULE entries MAIN MAIN 0 mainMain 1 bint Main 501 lazy_int Data.ByteString.Nums.Careless.Int 41211385 digitize Data.ByteString.Nums.Careless.Int 21210408 The number of entries to `lazy_int` is puzzling. Eugene's `bint` is called for each line of the file -- once for the header and then 5 million times for each of the integers. (There are two numbers on the first line but Eugene's program only uses `k` so `bint` is only actually entered once.) However, `bint` just calls my `int` and `int` calls `lazy_int` so why are there 41 million plus entries of `lazy_int`? -- Jason Dusek spoj-eugene-prof-opt-bang-acc-scc-dfold.prof Description: Binary data ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Can't get my head round monad transformers
Maybe the chapter on monad transformers in RWH can help? http://book.realworldhaskell.org/read/monad-transformers.html On Tue, Sep 1, 2009 at 7:34 PM, Colin Adamscolinpaulad...@googlemail.com wrote: I'm trying to add a state monad onto the IO monad for use in a happstack application. I thought this should involve using StateT and Happstack.Server.SimpleHTTP.simpleHTTP', but I can't figure out how to plumb it all together. Any tips will be welcome. -- Colin Adams Preston, Lancashire, ENGLAND ___ 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] Cleaning up stable names?
Have you considered computing a hash of the scene graph description (compositionally)? This gives you all the same advantages, including re-using shared subparts of the graph (which would be unlikely to stay around for very long as stable names). Luke On Tue, Sep 1, 2009 at 11:02 AM, Peter Verswyvelenbugf...@gmail.com wrote: I was planning to use them for caching OpenGL display lists and render targets. These are generated from a pure scene graph description, and if the same description is handed over to the render engine, it would just reuse the previously cached OpenGL object. I can of course just embedding and IORef inside the pure structures, that would also work I guess On Tue, Sep 1, 2009 at 6:09 PM, Job Vranishjvran...@gmail.com wrote: Well usually, when I've used stable names, I've just used them to check if things are the same, and then thrown them away. So no chance for a space leak. It's usually unsafe to keep stable names around for very long as they can lose their ability to tell if two things are the same (if this surprises you, you should carefully reread http://haskell.org/ghc/docs/latest/html/libraries/base/System-Mem-StableName.html#v%3AmakeStableName ). Out of curiosity, how are you planning on using them? - Job On Tue, Sep 1, 2009 at 3:40 AM, Peter Verswyvelen bugf...@gmail.com wrote: but without that function, stable names are not that useful I guess? they would cause a space leak? On Mon, Aug 31, 2009 at 10:59 PM, Job Vranish jvran...@gmail.com wrote: I also would like a isStableNameTargetAlive function. Though if you had such a function then you probably _could_ make a deRefStableName function, which, since there isn't one, probably means that such a function would be hard to make. - Job On Sun, Aug 30, 2009 at 4:54 PM, Peter Verswyvelen bugf...@gmail.com wrote: From the documentation, I don't think I grasp how stable names work. From the docs: There is no deRefStableName operation. You can't get back from a stable name to the original Haskell object. The reason for this is that the existence of a stable name for an object does not guarantee the existence of the object itself; it can still be garbage collected. From this I can conclude that stable names behave a bit like weak pointers. However, suppose I have a hash table of these stable names. How can I remove the redundant stable names from the table? I mean removing stable names that refer to an object that is garbage collected? I don't see any function for checking that (e.g. isStableNameTargetAlive or something) Thanks, Peter ___ 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 ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Cleaning up stable names?
Yes that is possible too. But it feels obvious to first lookup by stable name. I had no idea stable names were so volatile. They should be called unstable names then :-) In .NET it is possible to assign an identifier to an object, and that identifier will always be the same for the same object, no matter where to garbage collectors moves the object in memory. For Haskell, at first sight it would feel natural to have something like that too. On Tue, Sep 1, 2009 at 8:31 PM, Luke Palmerlrpal...@gmail.com wrote: Have you considered computing a hash of the scene graph description (compositionally)? This gives you all the same advantages, including re-using shared subparts of the graph (which would be unlikely to stay around for very long as stable names). Luke On Tue, Sep 1, 2009 at 11:02 AM, Peter Verswyvelenbugf...@gmail.com wrote: I was planning to use them for caching OpenGL display lists and render targets. These are generated from a pure scene graph description, and if the same description is handed over to the render engine, it would just reuse the previously cached OpenGL object. I can of course just embedding and IORef inside the pure structures, that would also work I guess On Tue, Sep 1, 2009 at 6:09 PM, Job Vranishjvran...@gmail.com wrote: Well usually, when I've used stable names, I've just used them to check if things are the same, and then thrown them away. So no chance for a space leak. It's usually unsafe to keep stable names around for very long as they can lose their ability to tell if two things are the same (if this surprises you, you should carefully reread http://haskell.org/ghc/docs/latest/html/libraries/base/System-Mem-StableName.html#v%3AmakeStableName ). Out of curiosity, how are you planning on using them? - Job On Tue, Sep 1, 2009 at 3:40 AM, Peter Verswyvelen bugf...@gmail.com wrote: but without that function, stable names are not that useful I guess? they would cause a space leak? On Mon, Aug 31, 2009 at 10:59 PM, Job Vranish jvran...@gmail.com wrote: I also would like a isStableNameTargetAlive function. Though if you had such a function then you probably _could_ make a deRefStableName function, which, since there isn't one, probably means that such a function would be hard to make. - Job On Sun, Aug 30, 2009 at 4:54 PM, Peter Verswyvelen bugf...@gmail.com wrote: From the documentation, I don't think I grasp how stable names work. From the docs: There is no deRefStableName operation. You can't get back from a stable name to the original Haskell object. The reason for this is that the existence of a stable name for an object does not guarantee the existence of the object itself; it can still be garbage collected. From this I can conclude that stable names behave a bit like weak pointers. However, suppose I have a hash table of these stable names. How can I remove the redundant stable names from the table? I mean removing stable names that refer to an object that is garbage collected? I don't see any function for checking that (e.g. isStableNameTargetAlive or something) Thanks, Peter ___ 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 ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Question about Lazy.IO
2009/09/01 Luke Palmer lrpal...@gmail.com: I have now tried twice to explain why this is possible more deeply in terms of domain theory, but it has bloated the message to three times this size. If you would like me to include my response, I'd be more than happy to, but it does get rather technical. I'm sure I'm not alone when I say I'd like to see the longer, more technical response. -- Jason Dusek ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Slow IO?
I've uploaded a new version of bytestring-nums that, while still slower than the fast/custom codes, allows Eugene's earlier program to a little more than 20% faster than it did before. It no longer handles spurious characters in the input by skipping over them (this is probably not a common requirement, anyways). http://hackage.haskell.org/package/bytestring-nums-0.3.0 I suspect that splitting the string into pieces and then mapping the parser over the pieces will never be faster than an all-in-one parser/tester/incrementer like the fast programs have. -- Jason Dusek ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Cleaning up stable names?
On Sep 1, 2009, at 14:57 , Peter Verswyvelen wrote: In .NET it is possible to assign an identifier to an object, and that identifier will always be the same for the same object, no matter where to garbage collectors moves the object in memory. For Haskell, at first sight it would feel natural to have something like that too. Hm. I'd think such names would have to live in a monad (which then leads you to either Reader or ST, I think). -- brandon s. allbery [solaris,freebsd,perl,pugs,haskell] allb...@kf8nh.com system administrator [openafs,heimdal,too many hats] allb...@ece.cmu.edu electrical and computer engineering, carnegie mellon universityKF8NH PGP.sig Description: This is a digitally signed message part ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] How to understand the 'forall' ?
data Branch tok st a = forall b. Branch (PermParser tok st (b - a)) (GenParser tok st b) please shed a light on me, thanks! -- View this message in context: http://www.nabble.com/How-to-understand-the-%27forall%27---tp25250783p25250783.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Question about Lazy.IO
I'm sure I'm not alone when I say I'd like to see the longer, more technical response. No, you aren't. Please `flood in' :-) -- ~dsouza yahoo!im: paravinicius gpg key fingerprint: 71B8 CE21 3A6E F894 5B1B 9ECE F88E 067F E891 651E ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Re: Snow Leopard Breaks GHC
On Tue, Sep 1, 2009 at 3:14 AM, Christian Maederchristian.mae...@dfki.de wrote: It seems, bootstrapping cabal went wrong. Does http://hackage.haskell.org/platform/2009.2.0.2/haskell-platform-2009.2.0.2-i386.dmg work? Installing the Haskell Platform package, combined with adding the previously mentioned options to /usr/bin/ghc (-optc-m32 -opta-m32 -optl-m32), seems to have done the trick. Thank you! ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] How to understand the 'forall' ?
This means that for any type 'b' you can construct a value of type 'Branch tok st a' by passing to Branch an argument of type '(PermParser tok st (b - a))' and 'GenParser tok st b'. This also means that when you're given a value of type Branch tok st a, you don't know what that 'b' type was; the only thing you know is that the 'b' in 'b - a' in the first argument of Branch is the same as the 'b' in 'GenParser tok st b'. See also: the haskellwiki page on existential types. 2009/9/2 zaxis z_a...@163.com: data Branch tok st a = forall b. Branch (PermParser tok st (b - a)) (GenParser tok st b) please shed a light on me, thanks! -- View this message in context: http://www.nabble.com/How-to-understand-the-%27forall%27---tp25250783p25250783.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe -- Eugene Kirpichov Web IR developer, market.yandex.ru ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Slow IO?
On Tue, 2009-09-01 at 08:45 +0400, Eugene Kirpichov wrote: Hm, on my machine Don's code has exactly the same performance my code above. That's strange. Also, replacing the 'test' and 'parse' functions with this one add :: Int - Int - S.ByteString - Int add k i s = fst $ S.foldl' f (i, 0) s where f (!i, !n) '\n' | n`divisibleBy`k = (i+1, 0) | otherwise = (i, 0) f (!i, !n) w = (i, 10*n+ord w-ord '0') increases performance by another 15% (0.675s vs 0.790s) On my system I get a 50% slowdown using this add function! I guess is just shows that benchmarking code on one single CPU/memory/OS/ghc combination does not give results that apply widely. I'm using: AMD Athlon X2 4800 2GB memory Linux (Fedora 11, 64-bit version) ghc 6.10.3 Steve ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Slow IO?
2009/9/2 Steve stevech1...@yahoo.com.au: On Tue, 2009-09-01 at 08:45 +0400, Eugene Kirpichov wrote: Hm, on my machine Don's code has exactly the same performance my code above. That's strange. Also, replacing the 'test' and 'parse' functions with this one add :: Int - Int - S.ByteString - Int add k i s = fst $ S.foldl' f (i, 0) s where f (!i, !n) '\n' | n`divisibleBy`k = (i+1, 0) | otherwise = (i, 0) f (!i, !n) w = (i, 10*n+ord w-ord '0') increases performance by another 15% (0.675s vs 0.790s) On my system I get a 50% slowdown using this add function! I guess is just shows that benchmarking code on one single CPU/memory/OS/ghc combination does not give results that apply widely. I'm using: AMD Athlon X2 4800 2GB memory Linux (Fedora 11, 64-bit version) ghc 6.10.3 I've got a Centrino Duo 2000 (I'm on a notebook), Ubuntu 9.04 and ghc 6.10.2. However, we have not set up on what exact input file we're using :) I'm using one where it is written 1000 3 and then 1000 lines of 9 follow. Also, I wonder what one'd get if one compiled this program with jhc, but I don't know whether jhc is able to compile Data.ByteString. Steve -- Eugene Kirpichov Web IR developer, market.yandex.ru ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Slow IO?
On Wed, 2009-09-02 at 11:55 +0800, Steve wrote:
On Tue, 2009-09-01 at 08:45 +0400, Eugene Kirpichov wrote:
Hm, on my machine Don's code has exactly the same performance my code above.
That's strange.
Also, replacing the 'test' and 'parse' functions with this one
add :: Int - Int - S.ByteString - Int
add k i s = fst $ S.foldl' f (i, 0) s
where f (!i, !n) '\n' | n`divisibleBy`k = (i+1, 0)
| otherwise = (i, 0)
f (!i, !n) w = (i, 10*n+ord w-ord '0')
increases performance by another 15% (0.675s vs 0.790s)
On my system I get a 50% slowdown using this add function!
I guess is just shows that benchmarking code on one single
CPU/memory/OS/ghc combination does not give results that apply widely.
I'm using:
AMD Athlon X2 4800
2GB memory
Linux (Fedora 11, 64-bit version)
ghc 6.10.3
I should have also said that the test method and test data is important
too.
This is what I have been using:
$ time ./0450 0450.input.data
and looking at the 'real' value.
The file 0450.input.data is generated with a Python script:
#!/usr/bin/env python
'''
generate a data file for problem 0450
'''
from __future__ import division# new in 2.2, redundant in 3.0
from __future__ import absolute_import # new in 2.5, redundant in
2.7/3.0
from __future__ import print_function # new in 2.6, redundant in 3.0
import io
import random
inFile = '0450.input.data'
#n, k, tiMax = 10**6, 3, 10**9
n, k, tiMax = 10**7, 3, 10**9
with io.open(inFile, 'wb') as f:
f.write('%d %d\n' % (n, k))
for i in xrange(n):
ti = random.randint(1, tiMax)
f.write('%d\n' % (ti,))
Steve
___
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Slow IO?
On 02/09/2009, at 2:26 PM, Eugene Kirpichov wrote: I've got a Centrino Duo 2000 (I'm on a notebook), Ubuntu 9.04 and ghc 6.10.2. However, we have not set up on what exact input file we're using :) I'm using one where it is written 1000 3 and then 1000 lines of 9 follow. Also, I wonder what one'd get if one compiled this program with jhc, but I don't know whether jhc is able to compile Data.ByteString. It couldn't last time I tried - choked on some INLINE pragmas. Might not be a massive job, but there aren't enough hours in the day... mark ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] How to understand the 'forall' ?
Isnot it clear without the 'forall' ? data Branch tok st a = Branch (PermParser tok st (b - a)) (GenParser tok st b) thanks! jkff wrote: This means that for any type 'b' you can construct a value of type 'Branch tok st a' by passing to Branch an argument of type '(PermParser tok st (b - a))' and 'GenParser tok st b'. This also means that when you're given a value of type Branch tok st a, you don't know what that 'b' type was; the only thing you know is that the 'b' in 'b - a' in the first argument of Branch is the same as the 'b' in 'GenParser tok st b'. See also: the haskellwiki page on existential types. 2009/9/2 zaxis z_a...@163.com: data Branch tok st a = forall b. Branch (PermParser tok st (b - a)) (GenParser tok st b) please shed a light on me, thanks! -- View this message in context: http://www.nabble.com/How-to-understand-the-%27forall%27---tp25250783p25250783.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe -- Eugene Kirpichov Web IR developer, market.yandex.ru ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe -- View this message in context: http://www.nabble.com/How-to-understand-the-%27forall%27---tp25250783p25251783.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] How to understand the 'forall' ?
You can also look at it in terms of the deconstruction operation on this data type. For example, lists: data List a = Nil | Cons a (List a) case_list :: r - (a - List a - r) - List a - r case_list nil_case cons_case Nil = nil_case case_list nil_case cons_case (Cons x xs) = cons_case x xs Now, given the Branch type: data Branch tok st a = forall b. Branch (PermParser tok st (b - a)) (GenParser tok st b) what is the type of its case construct? Here's the answer: case_branch :: (forall b. PermParser tok st (b - a) - GenParser tok st b - r) - Branch tok st a - r case_branch k (Branch pparser gparser) = k pparser gparser Notice the higher-rank type on the argument k; it has to be able to accept *any* type b, because the constructor hid an object of *some* type which is no longer known. So unless you can accept *any* b, you can't operate on this object. -- ryan ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] How to understand the 'forall' ?
2009/9/2 zaxis z_a...@163.com: Isnot it clear without the 'forall' ? data Branch tok st a = Branch (PermParser tok st (b - a)) (GenParser tok st b) thanks! The situation is not so simple. Consider, for example, a procedure that takes as input a *generic* sorting algorithm: sortThem :: (forall a. Ord a = [a] - [a]) - [Int] - [String] - ([Int], [String]) sortThem sortAlgo ints strings = (sortAlgo ints, sortAlgo strings) Here you can't omit the 'forall' because if you do, then inside the body of the sortThem function the 'a' type variable has a fixed value and can't be both Int and String! This is a somewhat esoteric example, but one could consider, for instance, a procedure that takes as input some trees of different types and a generic tree traversal algorithm. Well, have a look at the haskellwiki page, there's a pile of examples :) P.S. I tried to write up the difference between datatype and function declarations in this respect, but my explanations turned into a mess, so I erased them in the hope that someone will explain it better than me. jkff wrote: This means that for any type 'b' you can construct a value of type 'Branch tok st a' by passing to Branch an argument of type '(PermParser tok st (b - a))' and 'GenParser tok st b'. This also means that when you're given a value of type Branch tok st a, you don't know what that 'b' type was; the only thing you know is that the 'b' in 'b - a' in the first argument of Branch is the same as the 'b' in 'GenParser tok st b'. See also: the haskellwiki page on existential types. 2009/9/2 zaxis z_a...@163.com: data Branch tok st a = forall b. Branch (PermParser tok st (b - a)) (GenParser tok st b) please shed a light on me, thanks! -- View this message in context: http://www.nabble.com/How-to-understand-the-%27forall%27---tp25250783p25250783.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe -- Eugene Kirpichov Web IR developer, market.yandex.ru ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe -- View this message in context: http://www.nabble.com/How-to-understand-the-%27forall%27---tp25250783p25251783.html Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe -- Eugene Kirpichov Web IR developer, market.yandex.ru ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
