- Haskell-Cafe - Haskell.org

Re: IO monad and lazy evaluation
by Hal Daume III 22 May '03

22 May '03

Moved to haskell-cafe... > Thanks for your comments. (I had overlooked "readFile", which certainly > looks safer. I'm not yet fully used to the distribution of functions > between standard libraries and the Prelude. As for my 'hSafeGetContents', > I realized exactly your point about using 'id' too late, after I'd sent my > email. :-( ) I've done this many many times...too often do I find myself in the state of "open mouth, insert foot." People tend to be pretty forgiving though. > The broader debate I was trying to evoke was: how, as a programmer using a > combination of monads and non-strict functions, can I be confident that > there aren't hidden interactions waiting to bite? Well, I think it's more the combination of *strict* monads and non-strict functions that causes trouble. In fact, as pointed out before, lazy IO uses unsafeInterleaveIO to work. I think the rationale for lazy IO is clear. Without lazy IO, the following program, when run on a 2gb file, will allocate 2gb of memory: > main = interact (map toUpper) whereas, with lazy IO, it will run in constant (small) space. I think the point of lazy IO (and people who designed it can probably speak better to this) is to allow people to continue to write pure functions. That is, without lazy IO, the above would have to be written something like: > main = do > b <- hIsEOF stdin > when (not b) $ do > c <- hGetChar stdin > putChar (toUpper c) > main or something like that. obviously in *this case*, we would prefer to use the pure function map. However, in order to use map, we need to only read in characters as they are demanded, which means we need lazy IO. I think that what it really boils down to is you need to ask yourself: am I benefiting by using lazy IO. In the 'interact (map toUpper)' case, clearly you are, so you should probably continue to you it. However, in a case where you know you're going to read the whole file and store it in some data structure and need to simultaneously use data from the beginning and end of the file, then you're probably not benefiting from lazy IO. It is somewhat unfortunate that the IO library doesn't contain strict versions of all the operations, but they're easy enough to write with DeepSeq in your pocket. > Using the hOpen ... hClose vs readFile case as an example, is the > difficulty here to do with the interleaving of multiple statements that > modify some state with non-strict functions that return results based on > some particular instance of that state? Is this really a fundamental mismatch? Yes, this is precisely the problem. Using hGetContents implicitly uses unsafeInterleaveIO, which basically only performs IO as demanded. When you then say 'hClose h', none of the input has been demanded, so nothing gets read. Ever. Someone (one of the Simons, perhaps) could comment on this, but it seems that the primary problem with using readFile is the open handles issue, which boils down to the definition of hGetContents: > hGetContents :: Handle -> IO String > hGetContents handle = > withHandle "hGetContents" handle $ \handle_ -> > case haType handle_ of > ClosedHandle -> ioe_closedHandle > SemiClosedHandle -> ioe_closedHandle > AppendHandle -> ioe_notReadable > WriteHandle -> ioe_notReadable > _ -> do xs <- lazyRead handle > return (handle_{ haType=SemiClosedHandle}, xs ) If this were changed to actually close the handle, rather than leaving it in a semi closed state (call this new function hGetContentsAndClose or something) and then readFileAndClose were implemented based on this, would this solve the problem? > I read recently that other functional languages (ML?) with strict > evaluation semantics can be used to mimic lazy evaluation, but saw no clue > how that would work. I'm wondering if any insight might be gained by > looking at how such mechanisms might interact with functional > state-transformer mechanisms? The basic idea is to do something like: > data Lazy a = > Value a (Lazy a) > Deferred (() -> Lazy a) or something, and then you provide a function: > force (Value a l) = Value a l > force (Deferred f) = force (f ()) You can then force values as you need them; presumably in strict evaluation semantics, 'f ()' would be evaluated only at the point that it's forced. This isn't really relevant, though, I don't think, since it's sort of the opposite of what we want to be doing... - Hal

1 0

RE: getpid() or something similar
by Markus.Schnell＠infineon.com 22 May '03

22 May '03

You can't guarantee that another process didn't happen to write a file with just that name, pid or not. Markus > -----Original Message----- > From: Peter Simons [mailto:simons@cryp.to] > Sent: Thursday, May 22, 2003 12:06 PM > To: haskell-cafe(a)haskell.org > Subject: Re: getpid() or something similar > > > Simon Marlow writes: > > > Even using a ProcessID doesn't guarnatee uniqueness [...] > > Why not? If I use the file only temporarily (it is gone once the > process terminates), something like /tmp/foo.<pid> will be unique, for > all I can tell. > > > > The right way to do this is to try to open it for writing, and try > > a different name if the open fails. > > Unfortunately, this approach features a race condition because POSIX > has no notion of mandatory file locking. I'd really rather avoid this, > if I can. > > Is there no other way? And one that works across different compilers? > > Peter > > _______________________________________________ > Haskell-Cafe mailing list > Haskell-Cafe(a)haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe >

1 0

getpid() or something similar
by Peter Simons 22 May '03

22 May '03

Please pardon me if it turns out I'm just blind, but is there any function I can use to get some "unique" information to use for temporary file names and the like? A function that will tell me my process ID, for instance? I looked through Hugs and GHC, yet couldn't find anything but Data.Unique, and that looks like big-time overkill for the purpose. Any pointers are appreciated! Peter

3 3

Socratic method and Imperativicles
by Steffen Mazanek 21 May '03

21 May '03

Hello, I just have read Platon's Euthyphron and was fascinated by the maieutics of Socrates. So I want to start the little project "Imperativicles", which should result in a dialog similar to http://web.mit.edu/kerberos/www/dialogue.html. Thereby Socrates confuses an impassionate imperative programmer which thinks in "lines of code" exclusively and points him to the beauty of functional programming. But there is one problem. I cannot express these important, well-known phrases of Socrates in English. So if anybody wants to contribute I would be happy. All ideas, e.g. argumentations, would be helpful and greatly appreciated. The "project" page is the HaWiKi: http://haskell.org/hawiki/SocraticMethod Best regards, Steffen -- Steffen Mazanek #Haskell,that's where I just curry until fail,unwords www.steffen-mazanek.de#any error,drop all undefined,maybe break,otherwise in #sequence span isControl and take max $,id: (d:[])

2 1

Help: writing ffi bindings to a C library
by Bayley, Alistair 21 May '03

21 May '03

I want to write a set of bindings to Oracle's OCI library (a low-level C library for doing stuff to the Oracle database). Before I get started, has anyone done this already? (I don't think so, but I'd better ask.) My goal is to start with a simple set of low-level bindings, and build this up to higher-level database abstractions, and to hopefully port HaskellDB (if it's possible). But to start with I simply want to create a set of bindings for the OCI functions and get them working. Trouble is, I have no idea how to start writing a FFI binding. So far I've read: - Haskell 98 FFI Addendum (RC8) : a reference, but not a tutorial. - http://reid-consulting-uk.ltd.uk/docs/ffi.html : incomplete - http://haskell.org/ghc/docs/latest/html/users_guide/hsc2hs.html : is this for writing C that calls Haskell functions? Can anybody point me to a tutorial, or some illustrative examples? At the moment I'm trying to avoid the various ffi tools, and just use the basic ffi capabilities in GHC, so I can get an idea of how it all works. When I'm more comfortable with ffi then I'll think about using the various tools to speed things up. I'm not sure how to treat structures used by the C library. For example, a sample C program might have: int main () { sword rc; OCIEnv *envhp = NULL; OCIServer *srvhp = NULL; OCIError *errhp = NULL; OCISvcCtx *svchp = NULL; OCISession *usrhp = NULL; OCIStmt *stmtp = NULL; ... rc = OCIEnvCreate(&envhp, OCI_DEFAULT, 0, 0, 0, 0, 0, 0); rc = OCIHandleAlloc(envhp, (dvoid**) &errhp, OCI_HTYPE_ERROR, 0, 0); rc = OCIHandleAlloc(envhp, (dvoid**) &srvhp, OCI_HTYPE_SERVER, 0, 0); ... How do I handle values of OCIEnv, OCIServer, etc? I think I'm supposed to use Ptr, ForeignPtr, and Storable to reference them, but I don't see how to combine these types/classes (ForeignPtr takes values of type Ptr a; how do I construct a Ptr a? What type should "a" be?). ***************************************************************** The information in this email and in any attachments is confidential and intended solely for the attention and use of the named addressee(s). This information may be subject to legal professional or other privilege or may otherwise be protected by work product immunity or other legal rules. It must not be disclosed to any person without our authority. If you are not the intended recipient, or a person responsible for delivering it to the intended recipient, you are not authorised to and must not disclose, copy, distribute, or retain this message or any part of it. *****************************************************************

3 2

CC: Functional dependencies question
by oleg＠pobox.com 21 May '03

21 May '03

> hbar :: (Mult Energy Time t) => t > hbar = PhysicalUnit 6.62606876e-34 But what if we add to our library joule::Energy = PhysicalUnit 1.0 cal::Energy = PhysicalUnit 4.8 second::Time = PhysicalUnit 1.0 nanosec::Time = PhysicalUnit 1.0e-9 and scalarm:: Double -> (PhysicalUnit a1 b1 c1) -> (PhysicalUnit a1 b1 c1) x `scalarm` (PhysicalUnit y) = PhysicalUnit $ x * y then the user can write hbar = 6.62606876e-34 `scalarm` (joule `mult` second) or hbar = 6.62606876e-25 `scalarm` (joule `mult` nanosec) or hbar = (6.62606876e-25/4.8) `scalarm` (cal `mult` nanosec) Can't we? Or I'm missing something? GHC seems happy. We can probably define in the library transf:: (PhysicalUnit a1 b1 c1) -> (PhysicalUnit a1 b1 c1) -> Double transf (PhysicalUnit x) (PhysicalUnit y) = x/y and write hbar = (6.62606876e-25/(cal `transf` joule)) `scalarm` (cal `mult` nanosec) P.S. Just for the record: yet another way to introduce negative numbers is through an explicit constructor Neg. For example: http://pobox.com/~oleg/ftp/Computation/type-arithmetics.html (sorry, the code uses C++ templates. The idea is just the same however: a mere Prolog in a more verbose notation). The above example demonstrates division of signed Peano numbers. The Neg constructor becomes quite handy then.

3 2

please help me
by John Jonas Savimbi 20 May '03

20 May '03

From: John Jonas Savimbi Johannesburg, South Africa Email: jjsavimbi(a)phantomemail.com May 19TH, 2003. Dear Sir, It is my humble pleasure to write you this letter irrespective of the fact that you do not know me. However, I got your name through your country business directory here in my search for a reliable and trustworthy person that can assist me confidently. My name is Mr.John Jonas Savimbi. I am the son of Late Dr. Jonas Savimbi from Angola. I am 29 years old and I have got two younger sisters of same blood. Our mother died almost a year ago. We are all fighting for political asylum in South Africa. We cannot find a home in Angola anymore because of the war that our late father fought with the government of Angola until he died in that war. The government of Angola is now mad at us because our late father also wanted to be president and caused the people to fight the war for him because he was the president of UNITA political movement of Angola. So now we are in South Africa and we are suffering too much because we don't have money and the government of South Africa will not give jobs to asylum seekers. I am much concerned about my younger sisters who are finding it very difficult to manage in the situation. But I know that we should not suffer this way. I know that I must not let my younger sisters get into trouble because of money. So after consideration I am hereby begging for your assistance to help me to recover the sum of USD $29 Million which is my inheritance from my late mother. My late mother had taken this money with her to Spain in a diplomatic metal box marked (Precious Stones). Subsequently she lodged the metal box (As precious Stones) with a Securities and Valuables Protection company in Spain. I am supposed to be the beneficiary in the event of her death. However, all effort I made to go Spain to claim was refused by the Spainish embassy. And this was while it was still possible for me to travel. Now I cannot travel anywhere anymore. So in the situation that we are facing now, I must find somebody who can go to the Spain on my behalf. I understand that I will have to transfer my beneficiary rights unto you. I am willing to do so in the hope that you are a God fearing person who will not abandon us after you have the money. So from the bottom of my heart I am giving you 20% of the all the money. Once we have enough to sustain us sufficiently, you may be come our fund manager and invest the rest wisely for us. If you will be kind to assist us, please inform me urgently. Kind regards to you and your family. Sincerely, John Jonas savimbi

1 0

RE: Functional dependencies question
by Simon Peyton-Jones 19 May '03

19 May '03

Think of it like this. Should this be acceptable? f :: a -> a -> a f x y = x && y No, because the type (forall a. a->a->a) is plainly more general than the actual function. The type tempts you to think that you could write f 3 4 but you can't. But I suppose you could argue that the function definition should be allowed, but if you ever called (f 3 4) you should get an error "Int found where Bool expected". Your signature is bar :: (Foo Char t) => t and we know (from the instance decl) that t must be Bool. Any other call to bar would be illegal, just as any cal to 'f' above is illegal, except at type Bool. In general, it seems to make sense to reject definitions as early as possible, hence GHC and Hugs reject it. It is undoubtedly odd that adding an instance declaration makes a legal program illegal, and that might be a reason for relaxing the rule... but it's not entirely easy to implement such a relaxation. Simon | -----Original Message----- | From: haskell-cafe-admin(a)haskell.org [mailto:haskell-cafe-admin@haskell.org] On Behalf Of Andrew | J Bromage | Sent: 12 May 2003 06:10 | To: haskell-cafe(a)haskell.org | Subject: Functional dependencies question | | G'day everyone. | | Suppose I have code like this: | | class Foo a b | a -> b where | foo :: a -> b | | bar :: (Foo Char t) => t | bar = foo 'a' | | All well and good so far. Let's add an instance of Foo... | | instance Foo Char Bool where | foo c = isAlpha c | | Now neither GHC nor Hugs will allow this to compile, as the declared | type of bar is "too general". | | This seems intuitively the wrong behaviour. Either the original | program was incorrect (as there was no visible instance of Foo for | type Char) or the amended program is correct (since the type of bar | is no more general than it was in the first version). | | Thoughts? | | Cheers, | Andrew Bromage | _______________________________________________ | Haskell-Cafe mailing list | Haskell-Cafe(a)haskell.org | http://www.haskell.org/mailman/listinfo/haskell-cafe

4 3

n-term recurrences
by William Lee Irwin III 16 May '03

16 May '03

-- coefs starts with constant term, then coef of a(n-k) in k-term -- recurrence and goes up by indices -- vals starts with x0 and goes up by indices -- solves linear recurrences -- where xs !! (n+1) = head coefs + -- sum (zipWith (*) (tail coefs) [xs !! k | k <- [n - length coefs - 1..n]]) -- and the first few terms are given by xs !! n = vals !! n recurrence coefs vals | length coefs /= length vals + 1 = error "mismatched lengths of coefficients and values" | otherwise = vals ++ map coefSum ys where xs = recurrence coefs vals ys = transpose [drop k xs | k <- [0..length coefs - 1]] coefSum xs = head coefs + sum (zipWith (*) (tail coefs) xs) transpose xs = map head xs : transpose (map tail xs) -- This could be used, say, to generate the list of orthonormal -- polynomials generated by a three-term recurrence and then -- expand some function in a function series via Gaussian quadrature. -- e^(-x^2/2)H_n(x) or e^(-x^c/2)L^c_n(x) look especially attractive. -- OTOH that tactic may not be numerically useful, or useful only -- when the function series expansion is known a priori. -- It could also be a fancier way to generate Fibonacci numbers or -- Apery numbers or whatever. -- Not terribly deep, but it amused me for a few minutes to think of it. -- wli

1 0

ELearnChina Conference 2003
by Edwin Jones 16 May '03

16 May '03

Hi there, I hope you don't mind but I got your email address from your web site. You may be interested in coming to our conference [I think] which focuses on E-Learning, Knowledge Management, Training, Localisation and the like for the Chinese Market [biggest growth area for these types of product at the moment]. If you are interested please see http://www.chinaconferences.net The conference is in July in Edinburgh and there is now a special conference discount, which is available from the site [�400 GBP Discount on original rate]. Cheers, Edwin, Sales China Conferences Tel +44 131 440 9881 Fax +44 131 440 9882 edwin(a)chinaconferences.net

1 0