- Haskell-Cafe - Haskell.org

Re: [Haskell-cafe] Can you do everything without shared-memory concurrency?
by Richard A. O'Keefe 09 Sep '08

09 Sep '08

On 9 Sep 2008, at 8:15 am, Kyle Consalus wrote: > Anyway, for the time being I believe there are operations that can be > done with shared memory > that can't be done with message passing if we make "good performance" > a requirement. One of our people here has been working on Distributed Shared Memory for some years. It's a programming model where you write AS IF you had shared memory, but you really don't. He actually uses plain C with a library, but I've got a student project to wrap some syntax around that. typedef struct Foo { ... variables ... } Foo; typedef struct Bar { ... variables ... } Bar; shared Foo foo; shared Bar bar; shared (const *f = &foo) { ... here we have read access to f->variables ... } shared (*b = &bar) { ... here we have read-write access to b->variables ... } Underneath, it's message passing. When you get a view on a shared region, a copy is fetched from the cheapest source that has a current copy. When you release a write view, you become the holder of the only current copy. Compressed differences are sent around the local net. Zhiyi Huang's library is called VODCA. There's a plug-compatible version developed by his colleagues in China called Maotai, so the _same_ code can be run on a multicore system or on a network of workstations. The trick is to choose the chunk size of your problem so that computation and communication costs are in the right balance. This certainly seems to be adequate for numerical software, raytracing, game playing, ... One issue is that real shared memory comes at a price that most people don't know they are paying. We wouldn't need MOESI protocols or the related bus traffic if there were known to be no sharing, and one of the things that gets in the way of massive multicore is keeping caches coherent. No shared memory => no coherence problem => no extra bus traffic => faster. -- If stupidity were a crime, who'd 'scape hanging?

1 0

Re: [Haskell-cafe] Can you do everything without shared-memory concurrency?
by Richard A. O'Keefe 09 Sep '08

09 Sep '08

I think the demonstration is in Hoare's book on co-operating sequential processes, but if you have pure processes and message passing, you can simulate conventional variables. Here's an Erlang version: variable_loop(State) -> receive {ask,Sender} -> Sender!{self(),State}, variable_loop(State) ; {tell,New_State} -> variable_loop(New_State) end. new_variable(Initial_State) -> spawn(fun () -> variable_loop(Initial_State) end). fetch(Variable) -> Variable!{ask,self()}, receive {Variable,State} -> State end. store(Variable, New_State) -> Variable!{tell,New_State}. new_array(Size, Initial_State) -> list_to_tuple([new_variable(Initial_State) || Dummy <- lists:seq(1, Size)]). fetch(Array, Index) -> fetch(element(Index, Array)). store(Array, Index, New_State) -> store(element(Index, Array), New_State). The simulation of (shared) mutable variables and arrays in pure CSP is very similar. This pretty much _has_ to be possible in any language that has concurrent processes that can communicate in some fashion. There are also quite elementary ways to simulate locking. So we can solve any concurrent problem that, say, C++ with Intel's TBB library, or Java with java.util.concurrent, or any other such language can solve. We can even get ourselves just as far into extremely serious trouble as we can using those languages, it's just that we don't _have_ to. -- If stupidity were a crime, who'd 'scape hanging?

1 0

Re: building unix package on windows
by David Hotham 08 Sep '08

08 Sep '08

But the unix package isn't actually required to build HAppS on Windows. If you remove the dependency from the .cabal file (and also remove the -DUNIX build option) then you'll be fine.

1 0

monadic map on a Data.IntMap
by minh thu 08 Sep '08

08 Sep '08

Hi, Is there something like a fmapM_ ? In particular, I'd like to mapM_ a Data.IntMap instead of a List Thank you, Thu

4 4

Re: [Haskell-cafe] HXT from schema to data model
by Uwe Schmidt 08 Sep '08

08 Sep '08

Hello Pierre-Edouard, > Is there a way to generate a data model and a set of picklers from an XML (or > RelaxNG) Schema using the HXT tool box? the generation of a DTD out of the picklers is in an experimental stage. The generation of a RelaxNG Schema is technically possible, but it's an open project. It's on our TODO list for HXT. Validation with RelaxNG within HXT is complete. What is missinig is a complete datatype library for XML Schema Datatypes. The still missing part of the W3C XML Schema datatype library are the datatypes for date and time, all others are implemented. Cheers, Uwe -- Uwe Schmidt FH Wedel http://www.fh-wedel.de/~si/

1 0

Re: A problem with nested regions and higher-order?functions
by Chung-chieh Shan 08 Sep '08

08 Sep '08

Mario Bla??evi?? <mblazevic(a)stilo.com> wrote in article <23051.1216949255%241220747147(a)news.gmane.org> in gmane.comp.lang.haskell.cafe: > I'm trying to apply the nested regions (as in Lightweight Monadic > Regions by Oleg Kiselyov and Chung-chieh Shan) design pattern, if > that's the proper term, in hope to gain a bit more type safety in > this little library I'm working on (Streaming Component Combinators, > available on Hackage). I guess the problem is that I'm getting too > much type safety now, because I can't get the thing to compile. ... Hi! Thanks for your interest. It sounds like a promising application of region checking. I actually side with the type checker on this problem: > > type SingleHandler x y = forall r1s rs. Ancestor r1s rs => > > Handle r1s x -> Region rs y > > type DoubleHandler x y z = forall r1d r2d rd. (Ancestor r1d rd, Ancestor r2d rd) => > > Handle r1d x -> Handle r2d y -> Region rd z Here a SingleHandler is defined as an operation, in an arbitrary region on an arbitrary handle, that is valid as long as the region of the operation descends from the region of the handle. For example, reading a string from an arbitrary open file is a "SingleHandler File String". However, copying a string from an arbitrary open file to another fixed file is not a SingleHandler, because such an operation is valid only in a region that descends from the destination file handle's region! That's what GHC complained about as it checks your mapD: > > mapD :: (SingleHandler x z -> SingleHandler y z) > > -> DoubleHandler x w z -> DoubleHandler y w z > > mapD f d = \y w-> f (\x-> d x w) y So we really need to change the type of mapD if we want it to be accepted by a sound type system. The simplest way is to not require that mapD return a DoubleHandler (a binary operation that works in an arbitrary region): mapD :: ((Handle r1 x -> Region r z) -> (Handle r1 x -> Region r z)) -> ((Handle r1 x -> Handle r2 w -> Region r z) -> (Handle r1 x -> Handle r2 w -> Region r z)) mapD f d = \y w-> f (\x-> d x w) y This type is a special case of the type automatically inferred for mapD by Haskell if you omit the type signature. Haskell doesn't have any trouble type-checking higher-order functions, such as this mapD, but higher-rank types can call for manual annotations, as illustrated below. Another way to get mapD to type-check is to pass it (as its first argument) not a SingleHandler transformer but a transformer of operations that may work only in descendents of a given region r2. We want to write type H r2 x y = forall r1 r. (Ancestor r1 r, Ancestor r2 r) => Handle r1 x -> Region r y mapD :: (forall r2. H r2 x z -> H r2 y z) -> DoubleHandler x w z -> DoubleHandler y w z mapD f d = \y w-> f (\x-> d x w) y ("forall r2" takes scope over both "H r2 x z" and "H r2 y z" above) but GHC doesn't see that we intend to unify the "r2d" in the expansion of "DoubleHandler y w z" with the "r2" in this type signature of "mapD". So, we need to put an explicit type annotation on the use of "f" in the body of "mapD". To do so, we add {-# LANGUAGE ScopedTypeVariables #-} and expand the type synonym "DoubleHandler y w z" in the type signature of "mapD": mapD :: forall x y z w r1d r2d rd. (Ancestor r1d rd, Ancestor r2d rd) => (forall r2. H r2 x z -> H r2 y z) -> DoubleHandler x w z -> Handle r1d y -> Handle r2d w -> Region rd z mapD f d = \y w-> (f :: H r2d x z -> H r2d y z) (\x-> d x w) y Not so pretty anymore, but it does pass the type checker (GHC 6.8.2). -- Edit this signature at http://www.digitas.harvard.edu/cgi-bin/ken/sig ----- Ken Shan

1 0

experimental static blog engine in Haskell (file based, markdown syntax)
by jinjing 08 Sep '08

08 Sep '08

Hia, It's called Panda. It's pretty young, no theme, no tags, no comments, around 360 lines of code and uses Kibro to bootstrap. hosted on GitHub: http://github.com/nfjinjing/panda/ a quick demo at: http://jinjing.blog.easymic.com/ cheers, - jinjing

2 2

building unix package on windows
by jean-christophe mincke 08 Sep '08

08 Sep '08

Hello, I have installed GHC and cygwin on windows XP and I am trying to build the unix package (required to install HApps) ** When I use .../ghc/gcc as c compiler I receive the following error during the configure* D:\temp\haskell\unix-2.3.0.0>runhaskell Setup configure --ghc --prefix=D:\apps\h-lib Configuring unix-2.3.0.0... checking for gcc... gcc checking for C compiler default output file name... *configure: error: C compiler cannot create executables* See `config.log' for more details. and config.log says that : * ld: crt2.o: No such file: No such file or directory* ** When I use the gcc in cygwin : /usr/bin/gcc, the configure phase is ok but the build stop on the following error:* D:\temp\haskell\unix-2.3.0.0>runhaskell Setup build Preprocessing library unix-2.3.0.0... *In file included from System\Posix\DynamicLinker\Module.hsc:57:0: include/HsUnix.h:32:23: sys/times.h: No such file or directory* include/HsUnix.h:38:26: sys/resource.h: No such file or directory include/HsUnix.h:41:22: sys/wait.h: No such file or directory include/HsUnix.h:62:21: termios.h: No such file or directory include/HsUnix.h:65:25: sys/utsname.h: No such file or directory include/HsUnix.h:68:17: pwd.h: No such file or directory include/HsUnix.h:71:17: grp.h: No such file or directory include/HsUnix.h:81:17: pty.h: No such file or directory include/HsUnix.h:84:18: utmp.h: No such file or directory include/HsUnix.h:87:19: dlfcn.h: No such file or directory In file included from System\Posix\DynamicLinker\Module.hsc:57:0: include/HsUnix.h: In function `__hsunix_wifexited': include/HsUnix.h:103:0: warning: implicit declaration of function `WIFEXITED' include/HsUnix.h: In function `__hsunix_wexitstatus': include/HsUnix.h:104:0: warning: implicit declaration of function `WEXITSTATUS' include/HsUnix.h: In function `__hsunix_wifsignaled': include/HsUnix.h:105:0: warning: implicit declaration of function `WIFSIGNALED' include/HsUnix.h: In function `__hsunix_wtermsig': include/HsUnix.h:106:0: warning: implicit declaration of function `WTERMSIG' include/HsUnix.h: In function `__hsunix_wifstopped': include/HsUnix.h:107:0: warning: implicit declaration of function `WIFSTOPPED' include/HsUnix.h: In function `__hsunix_wstopsig': include/HsUnix.h:108:0: warning: implicit declaration of function `WSTOPSIG' include/HsUnix.h: In function `__hsunix_rtldDefault': include/HsUnix.h:115:0: error: `RTLD_DEFAULT' undeclared (first use in this function) include/HsUnix.h:115:0: error: (Each undeclared identifier is reported only once include/HsUnix.h:115:0: error: for each function it appears in.) include/HsUnix.h: In function `__hsunix_lstat': include/HsUnix.h:134:0: warning: implicit declaration of function `lstat' include/HsUnix.h: In function `__hsunix_mknod': include/HsUnix.h:140:0: warning: implicit declaration of function `mknod' compiling dist\build\System\Posix\DynamicLinker\Module_hsc_make.c failed *command was: d:\apps\ghc\bin\ghc.exe -c -package base-3.0.2.0 -package directory-1.0.0.1 -Iinclude dist\build\System\Pos ix\DynamicLinker\Module_hsc_make.c -o dist\build\System\Posix\DynamicLinker\Module_hsc_make.o* Could someone help me. Thank you J-C

2 1

Re: [Haskell] Top Level <-
by Adrian Hey 08 Sep '08

08 Sep '08

I have a feeling this is going to be a very long thread so I'm trying to go to Haskell cafe again (without mucking it up again). Derek Elkins wrote: > Haskell should be moving -toward- a capability-like model, not away from > it. Could you show how to implement Data.Random or Data.Unique using such a model, or any (preferably all) of the use cases identified can be implemented? Like what about implementing the socket API starting with nothing but primitives to peek/poke ethernet mac and dma controller registers? Why should Haskell should be moving -toward- a capability-like model and why does top level <- declarations take us away from it? Regards -- Adrian Hey

21 178

language proposal: ad-hoc overloading
by Ryan Ingram 07 Sep '08

07 Sep '08

The point of having a strongly typed language is so the compiler can do more work for you. But right now I do a lot of typing (pun not intended) to appease the compiler. Let me give you an example: module Prob where import qualified Data.Map as M ... newtype Prob p a = Prob { runProb :: [(a,p)] } combine :: (Num p, Ord a) => Prob p a -> Prob p a combine m = Prob $ M.assocs $ foldl' (flip $ uncurry $ M.insertWith (+)) M.empty $ runProb m Do you see it? All those "M." just seem dirty to me, especially because the compiler should be able to deduce them from the types of the arguments. My proposal is to allow "ad-hoc" overloading of names; if a name is ambiguous in a scope, attempt to type-check the expression against each name. It is only an error if type-checking against all names fails. If type-checking succeeds for more than one then the expression is ambiguous and this is also an error. Pros: shorter code, less busywork to please the compiler Cons: potentially exponential compile time? Any thoughts? -- ryan

16 25