- Glasgow-haskell-users

Re: Socket library ghc 5.02.1
by Sven Eric Panitz 28 Nov '01

28 Nov '01

From: "Sigbjorn Finne" <sof(a)galois.com> Cc: <glasgow-haskell-users(a)haskell.org> Date: Tue, 27 Nov 2001 09:30:04 -0800 > > Conclusion: you're hosed with ghc-5.02.1 and its socket libs under > Win32. Sorry. > If you don't mind getting your hands a (little) bit dirty, here's a story that will work ghc-5.02.1: * edit SocketPrim.hi (and SocketPrim.p_hi), to instead of saying "Socket" in its __export section it says "Socket{MkSocket}" (you'll find the .hi file in imports/net/ inside your 5.02.1 tree). * compile up the attached NetExtra.hs as follows: foo$ ghc -c NetExtra.hs -fvia-C -fglasgow-exts -package net * import and include NetExtra with your socket code, e.g., Thanks Danke Mercie, this works (although I did not fully understand, what I was doing). Sven Eric

1 0

Casting dynamic values
by George Russell 28 Nov '01

28 Nov '01

It would occasionally be nice to have a function cast :: (Typeable a,Typeable b) => a -> Maybe b which returns Just a if a and b have the same type, and Nothing otherwise. This may seem rather a curious need, but it arises with existential types; if you have data A = forall a . (context) => A a (context including Typeable a) then this allows you to getB :: (Typeable b) => A -> Maybe b getB (A a) = cast a and so extract a value from A, if you can guess its type. Clearly we can implement cast = fromDynamic . toDyn My question is: is this the most efficient way of doing it, or is there a better way?

2 1

Re: Socket library ghc 5.02.1
by Sven Eric Panitz 27 Nov '01

27 Nov '01

From: "Sigbjorn Finne" <sof(a)galconn.com> Cc: <glasgow-haskell-users(a)haskell.org> References: <200111220959.KAA29614(a)softwareag.com> Date: Mon, 26 Nov 2001 12:30:47 -0800 "Sven Eric Panitz" <sep(a)softwareag.com> writes: > > It seems that the Socket library does still not work > with ghc 5.02.1. > I tried the simple test: > > > main = > > do > > d <- connectTo "localhost" (PortNumber 80) > > hPutStr d "GET / HTTP/1.0\n\n" > > hFlush d > > c <- hGetContents d > > putStr c > > On Windows2000 I get the known error: > > *** Exception: does not exist > Action: getProtocolByName > Reason: no such protocol entry > (You, of course, need to wrap up that code with Socket.withSocketDo to start up WinSock first). FYI, in case you're planning on doing socket programming with GHC-5.02 on a Win32 platform, stay away from using the higher-level Socket module, since its IO.Handle based view of sockets is just broken. Stick with the lower-level SocketPrim interface instead. Specifically, stay away from using the following: * Socket. connectTo * Socket.accept * Socket.sendTo * Socket. recvFrom * SocketPrim.socketToHandle --sigbjorn Thanks for the help and valuable information. I tried the following little test, which stays away from above functions: > module Main where > > import BSD > import SocketPrim > import Socket (withSocketsDo) > > main = > Socket.withSocketsDo > (do > protNum <- getProtocolNumber "tcp" > s <- socket AF_INET Stream protNum > hostAddr <- inet_addr "157.189.164.68" > let sAddr = (SockAddrInet (toEnum 8080) hostAddr) > connect s sAddr > i <- sendTo s "GET / HTTP/1.0\r\n\r\n" sAddr > (str,l,imsAddr) <- recvFrom s 1000 > putStr str > ) But something I seem to be doing wrong. During evaluation of 'recvFrom s 1000' I get the following error message (consistently for unix and windows): Fail: SocketPrim.hsc:241: Non-exhaustive patterns in case As the log of the webserver reveals the sendTo works fine. Any idea what could be my problem? Sven Eric

2 2

RE: (no subject)
by Yu Chen 26 Nov '01

26 Nov '01

hi, i've found the reason with someone's help. it is because i used "%code" instead of "safecode" in the gc file where i did callback. thanks! ---- chenyu ---------- Original Message ---------------------------------- From: "Simon Marlow" <simonmar(a)microsoft.com> Date: Mon, 26 Nov 2001 09:42:39 -0000 > >> i use green-card with ghc's FFI export to sth like >> >> Haskell -----> C ------> Haskell >> >> the program is just arithmatic operations. the program "seg fault" >> in the callback from C to Haskell. anyone whether it is a bug? > >There aren't any known bugs, but there are many known pitfalls. You'll >need to give us more information, at least: the code that causes the >crash (preferably a small example), the exact GHC command lines used to >compile it, your GHC version, your machine and OS version. > >Cheers, > Simon >

1 0

RE: GHC garbage collector
by Simon Marlow 26 Nov '01

26 Nov '01

> I have some code that uses FFI capabilities to associate > previously allocated > memory with a structure akin to a ByteArray and allows > Haskell land to read > and write to the C side of things. > > I use ForeignPtr and associate a destructor to allow the GHC's garbage > collector to clean up after it figures out that the structure > is no longer > being used. I know I am in funny territory. > > In some instances (when benchmarking) GHC takes its time > getting around to > garbage collecting. Can someone suggest hooks to encourage > GHC to clean up > things? Or better still functions to instruct GHC to garbage > collect specific > structures (by calling relevant destructors)? Also comments > on how safe this > whole business is. IOExts.performGC will cause a full garbage collection to be performed. If you want to be sure the finalizer has run to completion, you'll need to set up some synchronisation between the main thread and the finalizer using MVars. Finalizers have no special scheduling properties; they just run concurrently with the other threads in the system. Cheers, Simon

1 0

RE: (no subject)
by Simon Marlow 26 Nov '01

26 Nov '01

> i use green-card with ghc's FFI export to sth like > > Haskell -----> C ------> Haskell > > the program is just arithmatic operations. the program "seg fault" > in the callback from C to Haskell. anyone whether it is a bug? There aren't any known bugs, but there are many known pitfalls. You'll need to give us more information, at least: the code that causes the crash (preferably a small example), the exact GHC command lines used to compile it, your GHC version, your machine and OS version. Cheers, Simon

1 0

(no subject)
by Yu Chen 24 Nov '01

24 Nov '01

hi, there, i use green-card with ghc's FFI export to sth like Haskell -----> C ------> Haskell the program is just arithmatic operations. the program "seg fault" in the callback from C to Haskell. anyone whether it is a bug? TIA! ------ chenyu

1 0

GHC garbage collector
by Rahul Bhargava 24 Nov '01

24 Nov '01

Hi, I have some code that uses FFI capabilities to associate previously allocated memory with a structure akin to a ByteArray and allows Haskell land to read and write to the C side of things. I use ForeignPtr and associate a destructor to allow the GHC's garbage collector to clean up after it figures out that the structure is no longer being used. I know I am in funny territory. In some instances (when benchmarking) GHC takes its time getting around to garbage collecting. Can someone suggest hooks to encourage GHC to clean up things? Or better still functions to instruct GHC to garbage collect specific structures (by calling relevant destructors)? Also comments on how safe this whole business is. Before I launch myself into making obscene calls to GHC's internals, I should know what I am dealing with. Cheers Rahul

1 0

RE: Existentials and newtype
by Simon Peyton-Jones 23 Nov '01

23 Nov '01

| Summarizing: without single-field existentially quantified | constructors in newtypes either my code gets less readable or | I have to pay a run-time price. Is it possible to relax the | restriction on newtypes in combination with existentials? Yes, that's a fair point. Unfortunately, newtype existentials are tricky in GHC, because GHC's intermediate language is strongly typed. At the moment, existentials show up in that intermediate language solely mediated through data constructors. Newtypes have gone away by the time we are in the intermediate language. So there's nowhere to 'hang' the existential for a newtype. Would be easy to add existentials to the intermediate language? Unfortunately, not very, unless I misunderstand and someone can help me out. Corresponding to the types, we have terms to introduce and elminate the type: Forall type Big lambda e.g. /\a. \x::a -> e Type application e.g. f Int 4 Existential type what? Well, it's well understood what you need for existentials: a sort of pair for introduction and a sort of case for elimination. But I am deeply reluctant to add two new constructors to the Core language data type, which only has 8 constructors right now, and which deals quite happily with data-constructor-mediated existentials. Adding extra forms of Core would complicate every single compiler pass. I've written at some length because I want first class existentials too; see the paper that Mark and I wrote http://research.microsoft.com/~simonpj/papers/first-class-modules But I'm a bit stuck at the moment about how to slip them into Core without making a big heap of work. Simon

1 0

Re: hsc: Passing -1 as CString/Ptr a?
by Marcin 'Qrczak' Kowalczyk 22 Nov '01

22 Nov '01

Thu, 22 Nov 2001 13:57:32 +0100, Volker Stolz <vs(a)foldr.org> pisze: > Hi, I want to invoke the C-function 'dlsym', which has three > possible options for a parameter of type (void*): > > - NULL > - void* to a string, i.e. CString in Haskell > - RTLD_NEXT = (void *)(-1) You can obtain the value from an embedded C function. foreign import unsafe rtldNext :: Ptr a #def inline void *rtldNext (void) {return RTLD_NEXT;} -- __("< Marcin Kowalczyk * qrczak(a)knm.org.pl http://qrczak.ids.net.pl/ \__/ ^^ QRCZAK

1 0