- Glasgow-haskell-users

Re: Runtime importing of modules in GHC
by Andre Pang 27 May '02

27 May '02

On Mon, May 27, 2002 at 04:31:02PM +1000, Andre Pang wrote: > The concept: You want to have the equivalent of dlopen(); i.e. be > able to tell the Haskell runtime system to import a module and > call functions in that module. As an example, say there's > a function called adder: .. and now, my real reason for posting the previous message 8) Unfortunately, the big limitation of this is that we're going through the FFI to get things to work. This kinda sucks because that means that the exported functions have to stick to exporting one of the prim_types (Int, Char, Float, Double, Ptr, StablePtr, and a few others). So, is there some wizardry out there which allows using the RTS Linker to load up .o modules which are not created with the FFI? I've been experimenting with it, and I just get segfaults if I try to load the adder_closure, adder_entry, adder_fast1 or adder_info symbols. -- #ozone/algorithm <ozone(a)algorithm.com.au> - trust.in.love.to.save

1 0

Runtime importing of modules in GHC
by Andre Pang 27 May '02

27 May '02

Hi all, Manuel Chakravarty has helped me[1] with getting runtime importing of modules (a.k.a. dynamic loading) working in GHC on Linux. It should work on Windows, too. The concept: You want to have the equivalent of dlopen(); i.e. be able to tell the Haskell runtime system to import a module and call functions in that module. As an example, say there's a function called adder: module Adder1 where adder n = n + 2 which you want to call from another module, called, erm, "Main": module Main where main = do -- somehow import the Adder module let foo = adder 5 putStr (show foo ++ "\n") -- should print out 7 We've managed to get the concept working, and the actual code isn't all that complex. We compile the Linker.lhs file (from the compiler/ghci/ directory in the GHC source) so that we can get access to the functions in the Haskell Runtime System's linker. Then you can utilise the FFI to runtime-import the module that you want. Here's some example code (sorry to make this email so long): Adder1.hs: module Adder1 where foreign export adder :: (Int -> Int) adder n = n + 2 Adder2.hs: module Adder2 where foreign export adder :: (Int -> Int) adder n = n + 3 Main.hs: module Main where import Linker import Foreign import Maybe import System foreign import dynamic dynCallIntInt :: Ptr (Int -> Int) -> (Int -> Int) main :: IO () main = do -- initialise the runtime system linker initLinker -- Need to load these module so that resolveObjs later doesn't complain loadObj "/usr/lib/ghc-5.02.2/HSstd.o" loadObj "/usr/lib/ghc-5.02.2/HSstd_cbits.o" -- Load the module which we really want to load argv <- getArgs moduleName <- case argv of [args] -> return args _ -> do putStrLn "usage: ./a.out Foo," putStrLn " where Foo is the module name which contains the adder function" exitWith (ExitFailure 1) loadObj (moduleName ++ ".o") loadObj (moduleName ++ "_stub.o") putStr ("Loaded " ++ moduleName ++ "\n") -- Tell the linker to resolve the symbols in the newly-loaded object file resolvedObjs <- resolveObjs putStr ("Resolved Object Symbols: " ++ show resolvedObjs ++ "\n") -- Look up the function name we want to call foundSymbol <- lookupSymbol "adder" let justFoundSymbol = fromJust foundSymbol -- Call the function with appropriate parameters let result = dynCallIntInt justFoundSymbol 6 putStr (show result ++ "\n") -- Finished! putStr "Yay!\n" .. and the demonstration: 16:25(0) .../project/runtime_import% make ghc -fglasgow-exts -package lang -c `ghc --print-libdir`/HSrts.o Linker.lhs ghc -fglasgow-exts -package lang -c `ghc --print-libdir`/HSrts.o Adder1.hs ghc -fglasgow-exts -package lang -c `ghc --print-libdir`/HSrts.o Adder2.hs ghc -ldl -package lang -fglasgow-exts -fvia-c Main.hs Linker.o nice make 6.29s user 0.65s system 94% cpu 7.359 total 16:25(0) .../project/runtime_import% ./a.out Adder1 Loaded Adder1 Resolved Object Symbols: True 8 Yay! 16:25(0) .../project/runtime_import% ./a.out Adder2 Loaded Adder2 Resolved Object Symbols: True 9 Yay! 16:25(0) .../project/runtime_import% So it looks like it all works nicely :). The sample code's available at: http://www.algorithm.com.au/hacking/haskell/ if anybody wants to take a look at it. 1. Okay, he had about 95% of the insight, and I managed to piece the other 5% together :) -- #ozone/algorithm <ozone(a)algorithm.com.au> - trust.in.love.to.save

1 0

Medical Breakthrough... aging can be reversed with HGH
by medbrkthrgh77157533＠yahoo.com 25 May '02

25 May '02

There are three different types of HGH products. The confusion is that all three are advertised as if they were the same. The three types are: 1) --- Homeopathic HGH 2) --- Pre-cursor HGH 3) --- Real or synthetic HGH (delivered by injection or, by an oral spray method). Do you know differences? Call us and we'll explain them to you. Our toll free number is 1-888-621-7300 An HGH staff member is avialable 9 to 5 Pacific Time. If after hours, please leave you name and day and evening phone numbers. We will call you back in a no pressure, educational manner. If you are overseas call your long distance operator and ask to be connected to our phone number. We will call you back so we can pay for the long distance charges. For more information on HGH read on............ HAVE YOU HEARD OF HUMAN GROWTH HORMONE (HGH)??? Released by your own pituitary gland, HGH starts declining in your 20s, even more in your 30s and 40s, eventually resulting in the shrinkage of major organs -- plus, all other symptoms related to old age. IN THOUSANDS OF CLINICAL STUDIES, HGH HAS BEEN SHOWN TO ACCOMPLISH THE FOLLOWING: * Reduce Body Fat and Build Lean Muscle WITHOUT EXERCISE! * Enhance Sexual Performance * Remove Wrinkles and Cellulite * Lower Blood Pressure and Improve Cholesterol Profile * Improve Sleep, Vision and Memory * Restore Hair Color and Growth * Strengthen the Immune System * Increase Energy and Cardiac Output * Turn back your body's Biological Time Clock 10 - 20 years * Live Longer AND Stronger All natural and organic plant based FEEL 10 YEARS YOUNGER WITH ORAL SPRAY HGH. GUARANTEED We are the manufacturer and we sell directly to Doctors, Chiropractors, and consumers world wide the highest grade HGH Oral Spray available. With internet marketing, we are able to save advertising cost and pass those savings along to you. But you must act now. To receive more information call us now. TOLL FREE 1-888-621-7300 We must speak to you in person to qualify your usage. All of your questions will be addressed and answered in a friendly, no pressure manner. Our main purpose is to provide you with information so you can make an educated decision. For more information call 1-888-621-7300 If you are on line write down our phone number and call us when you can. Soon, you and your loved ones will be very glad you did. Read what people are saying: "The effects of 6 months of GH on lean body mass and fat were equivalent in magnitude to the changes incurred during 10-20 years of aging." Dr. Daniel Rudman, MD, New England Journal of Medicine. "Within four months, my body fat decreased form 30% down to 21%! I noticed my skin is more supple and my overall mental outlook improved significantly." D.W., New Jersey "We have been on the spray for just 3 weeks now, and besides the tremendous energy we both feel, my husbands allergies and spells of depression have lifted. I am healing extremely fast after an accident and have lost 7 lbs. without trying!" C.B., Flagstaff. AZ Thanks for reading our letter, The HGH Staff USA Division PS: The HGH Staff guarantees the highest quality and lowest price. We manufacture and ship directly to your door. Call us now 1-888-621-7300 ======= End of message ======== The following statement is provided to be in compliance with commercial email laws. If you do not wish to receive further mailings, please click reply and type remvoe in the subject box. Then click send. This message is in full compliance with U.S. Federal requirements for commercial email under bill S.1618 Title lll, Section 301, Paragraph (a)(2)(C) passed by the 105th U.S. Congress and is not considered SPAM since it includes a remove mechanism.* This message is not intended for residents in the states of CA, NC, NV, RI, TN, VA & WA. Screening of addresses has been done to the best of our technical ability. Call us now 1-888-621-7300 for your free HGH consultation. Thank you -- 1575332211111

1 0

RE: getting a grip on memory usage
by Simon Peyton-Jones 24 May '02

24 May '02

You surely need a seq in loop' (not loop). GHC can't see that it's strict, because it's overloaded. Something like | where loop' a pos | pos > high = a | | otherwise = a1 `seq` p1 `seq` loop' a1 p1 where a1 = f a pos p1 = p+1 (The seq on p1 simply avoids the creation of a thunk for p+1; the thunk will be evaluated on the next iteration, assuming '>' is strict, so it won't be a leak.) Or specialise loop to type Int (or whatever), which will help the strictness analyser no end. I'm not certain that this will do it, but if you don't do this you are in deep trouble. By 'eat memory' I assume you mean residency rather than just allocation Simon | -----Original Message----- | From: Hal Daume III [mailto:hdaume@ISI.EDU] | Sent: 22 May 2002 18:49 | To: GHC Users Mailing List | Subject: getting a grip on memory usage | | | so i have a function that is eating *tons* of memory. my | application is | clustering and i put everything in UArrays or Arrays. for | clustering 5 | data points with 4 features (a tiny tiny set) the program gobbles | 350mbs! most of the usage is coming from this distance function: | | dist dat Nothing pq@(p,q) = | (uncurry loop) (bounds xp) 0 | (\a i -> a + (sqr (xp!i - xq!i))) | where Vector xp = dat !!! p | Vector xq = dat !!! q | dist dat (Just (Vector w)) pq@(p,q) = | (uncurry loop) (bounds xp) 0 | (\a i -> a + ((sqr (w!i)) * (sqr (xp!i - xq!i)))) | where Vector xp = dat !!! p | Vector xq = dat !!! q | | Where the relevant definitions are: | | type Vector = Vector (UArray Int Double) | (!!!) = Array.(!) | | and loop is: | | loop :: (Num i, Ord i, Ix i) => i -> i -> a -> (a -> i -> a) -> a | loop low high a f = loop' a low | where loop' a pos | pos > high = a | | otherwise = loop' (f a pos) (pos+1) | | i cannot figure out why such a function would be eating so | much memory. i | even tried changing the function provided to loop with "a | `seq` ..." to | make sure we're not creating a huge thunk, but that didn't | help at all. | | but, according to my profiling: | | dist Main 380000 27.7 23.6 | 64.9 61.0 | sqr Main 2280000 1.6 7.9 | 1.6 7.9 | !!! Main 760000 4.7 0.0 | 4.7 0.0 | loop Loops 380000 30.8 29.5 | 30.8 29.5 | | i realize it's being entered a lot of times, and i can | understand that it | would use a bunch of time resources, but i don't understand | why it's using | so much space. | | any help would be appreciated... | | - Hal | | -- | Hal Daume III | | "Computer science is no more about computers | hdaume(a)isi.edu | than astronomy is about telescopes." -Dijkstra | www.isi.edu/~hdaume | | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users |

1 0

Re: List.sort
by Serge D. Mechveliani 24 May '02

24 May '02

> Just to let you know, Simon Marlow put the mergesort implementation into > the CVS libraries a bit over a week ago: > > http://cvs.haskell [..] Does this mean that next GHC release implements standard List.sort with the `merge' algorithm? ----------------- Serge Mechveliani mechvel(a)botik.ru

1 0

RE: LGPL and static linking of libgmp
by Simon Marlow 23 May '02

23 May '02

> I just re-read the GNU Lesser General Public License today - and I > discovered that it places some utterly strange (IMHO) restrictions on > executables statically linked with LGPLed libraries. > Unfortunately, GHC for Windows produces such executables by linking > in libgmp statically. Dynamically linking with libgmp would cause no > restrictions, but if it is statically linked, the LGPL says that > everyone distributing a program compiled with GHC has to 'Accompany > the work with the complete corresponding machine-readable source code > for the Library including whatever changes were used in the work > (which must be distributed under Sections 1 and 2 above); and, if the > work is an executable linked with the Library, with the complete > machine-readable "work that uses the Library", as object code and/or > source code, so that the user can modify the Library and then relink > to produce a modified executable containing the modified Library. (It > is understood that the user who changes the contents of definitions > files in the Library will not necessarily be able to recompile the > application to use the modified definitions.)' Yes, this is a pain. We really ought to distribute GMP as a DLL on Windows, and stop making static versions on Unix (the Solaris distribution of GHC generally includes a static libgmp, but most of the other dists don't). Better still, we could find an alternative to GMP - I don't know of one that is as fast & provides all the functionality that we use from GMP, though. Cheers, Simon

1 0

getting a grip on memory usage
by Hal Daume III 22 May '02

22 May '02

so i have a function that is eating *tons* of memory. my application is clustering and i put everything in UArrays or Arrays. for clustering 5 data points with 4 features (a tiny tiny set) the program gobbles 350mbs! most of the usage is coming from this distance function: dist dat Nothing pq@(p,q) = (uncurry loop) (bounds xp) 0 (\a i -> a + (sqr (xp!i - xq!i))) where Vector xp = dat !!! p Vector xq = dat !!! q dist dat (Just (Vector w)) pq@(p,q) = (uncurry loop) (bounds xp) 0 (\a i -> a + ((sqr (w!i)) * (sqr (xp!i - xq!i)))) where Vector xp = dat !!! p Vector xq = dat !!! q Where the relevant definitions are: type Vector = Vector (UArray Int Double) (!!!) = Array.(!) and loop is: loop :: (Num i, Ord i, Ix i) => i -> i -> a -> (a -> i -> a) -> a loop low high a f = loop' a low where loop' a pos | pos > high = a | otherwise = loop' (f a pos) (pos+1) i cannot figure out why such a function would be eating so much memory. i even tried changing the function provided to loop with "a `seq` ..." to make sure we're not creating a huge thunk, but that didn't help at all. but, according to my profiling: dist Main 380000 27.7 23.6 64.9 61.0 sqr Main 2280000 1.6 7.9 1.6 7.9 !!! Main 760000 4.7 0.0 4.7 0.0 loop Loops 380000 30.8 29.5 30.8 29.5 i realize it's being entered a lot of times, and i can understand that it would use a bunch of time resources, but i don't understand why it's using so much space. any help would be appreciated... - Hal -- Hal Daume III "Computer science is no more about computers | hdaume(a)isi.edu than astronomy is about telescopes." -Dijkstra | www.isi.edu/~hdaume

2 1

LGPL and static linking of libgmp
by Wolfgang Thaller 22 May '02

22 May '02

I just re-read the GNU Lesser General Public License today - and I discovered that it places some utterly strange (IMHO) restrictions on executables statically linked with LGPLed libraries. Unfortunately, GHC for Windows produces such executables by linking in libgmp statically. Dynamically linking with libgmp would cause no restrictions, but if it is statically linked, the LGPL says that everyone distributing a program compiled with GHC has to 'Accompany the work with the complete corresponding machine-readable source code for the Library including whatever changes were used in the work (which must be distributed under Sections 1 and 2 above); and, if the work is an executable linked with the Library, with the complete machine-readable "work that uses the Library", as object code and/or source code, so that the user can modify the Library and then relink to produce a modified executable containing the modified Library. (It is understood that the user who changes the contents of definitions files in the Library will not necessarily be able to recompile the application to use the modified definitions.)' I think it would be best to have future distributions of GHC for Windows include libgmp as a DLL. That's what I'll be doing for my future releases of GHC for MacOS X. Cheers, Wolfgang Thaller

1 0

RE: -fallow-overlapping-instances Request
by Simon Peyton-Jones 21 May '02

21 May '02

I don't understand all the implications of this. For example, what if we have an instance that doesn't match C () () now, but may do 'later' when a type variable has been instantiated. Overlapping instances are a terrific swamp and while they are interesting I don't think I'm going to spend a lot of time on them in the near future, I'm afraid. But perhaps someone else can.. Simon | -----Original Message----- | From: Ashley Yakeley [mailto:ashley@semantic.org] | Sent: 17 May 2002 10:02 | To: GHC List | Subject: -fallow-overlapping-instances Request | | | Currently -fallow-overlapping-instances only allows | overlapping instances | if one is a strict subset of the other. This is good | (determinate), but | sometimes you really need two instances that partially | overlap. It would | be nice if this could be disambiguated simply with another instance | declaration. For instance: | | class C a b; | instance C a (); | instance C () a; | instance C () (); | | As you can see, the first two instances partially overlap, | but the third | one disambiguates. I think it would be nice if GHC | -fallow-overlapping-instances allowed this. | | This would take some of the pain out of overlapping instance | resolution... | | -- | Ashley Yakeley, Seattle WA | | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-| haskell-users |

1 0

Re: -fallow-overlapping-instances Request
by Ashley Yakeley 18 May '02

18 May '02

At 2002-05-17 03:54, Johannes Waldmann wrote: >From a type-theoretic viewpoint, >instance declarations are relations between (sets of) trees >(= elements of the respective data types). >So one needs representations of such relations >with effective decidability of overlapping-ness, most-sepcific-ness and such. Yeah but unfortunately GHC doesn't look at the contexts of instance declarations. It has a straightforward but rather conservative concept of "overlapping" and "most specific". I'd like to see that fixed too, but that's a separate issue. -- Ashley Yakeley, Seattle WA

1 0