- Glasgow-haskell-users

Illegal instruction in GHC 6.5 generated code.
by Caio Marcelo 26 Jun '06

26 Jun '06

(If this is the wrong list for this, please tell me. I tried Glasgow-Haskell-Bugs, but message got hold in moderation) Hello, Using GHC 6.5 from darcs repo ("pulled" in 19/Jun), the following snippet of code seems to cause illegal instruction error on the generated code. The error disappears if I use the Dummy instance on top of Refeable. The code is from my bindings to Judy library, revision 10752 in subversion. http://svn.openfoundry.org/pugs/third-party/HsJudy/ http://svn.openfoundry.org/pugs/third-party/judy/ # original C lib code To test it you'll need to compile (./configure; make # but not necessarily install) the original library in third-party/judy/Judy-1.0.3. Back in HsJudy, change the Makefile to static link option. HsJudy$ make testmap HsJudy$ ./testmap Sorry, I tried to come up with an simpler instance of the problem but didn't found one. The program depend on WordPtr which isn't available in GHC 6.4.2, so I couldn't test it in stable version. It was tested on my amd64, linux 2.6.16 and in audreyt's machine (OSX I think). ---------8<------------ --class Dummy a --instance Dummy a --instance Dummy a => Refeable a where instance Refeable a where toRef a = do a' <- newStablePtr a return (ptrToWordPtr (castStablePtrToPtr a')) fromRef v = do a <- deRefStablePtr (castPtrToStablePtr (wordPtrToPtr v)) return a -- Don't need to StablePtr "simple" things like Int instance Refeable Int where toRef i = return $ toEnum i fromRef v = return $ fromEnum v ---------->8------------- Thanks in advance, -- Caio Marcelo

3 4

Compiling with -threaded under Solaris 10
by Florian Lorenzen 26 Jun '06

26 Jun '06

Hi all, I'm trying to run the following program on Solaris 10 with GHC 6.4.1 (6.4.2 does not really work under Solaris 10 right now). module Main where import IO import Control.Concurrent import System.Process main = do (inp, out, err, ph) <- runInteractiveProcess "cat" [] Nothing Nothing forkIO (do hPutStr inp "this\nis\na\ntest\n" hClose inp) forkIO (do r <- hGetContents out putStr r) waitForProcess ph I compile it with $ ghc -o try Try2.hs -threaded -lrt When running, I get $ ./try try: waitForProcess: interrupted (Interrupted system call) and the program aborts. Under Linux, everything is fine. Do I have to set some macros explicitly? And if so, which ones? Any help appreciated. Best regards, Florian

2 2

GHC driver behaviour switch
by kyra 23 Jun '06

23 Jun '06

With GHC 6.4.x I was able to use a ghc driver to compile .cpp sources. Now, with GHC HEAD I cannot. GHC HEAD now ALWAYS enforces "treat as .c" option. Let's compare "DriverPipeline.hs". GHC 6.4.x: let langopt | hcc = [ SysTools.Option "-x", SysTools.Option "c"] | otherwise = [ ] SysTools.runCc dflags ( -- force the C compiler to interpret this file as C when -- compiling .hc files, by adding the -x c option. -- Also useful for plain .c files, just in case GHC saw a -- -x c option. langopt ++ ... GHC HEAD: SysTools.runCc dflags ( -- force the C compiler to interpret this file as C when -- compiling .hc files, by adding the -x c option. -- Also useful for plain .c files, just in case GHC saw a -- -x c option. [ SysTools.Option "-x", SysTools.Option "c"] ++ ... This completely breaks my build infrastructure. Could it be reversed? Cheers, Kyra

3 4

FFI: number of worker threads?
by Li, Peng 22 Jun '06

22 Jun '06

Hello, The paper "Extending the Haskell FFI with Concurrency" mentioned the following in Section 6.3: "GHC's run-time system employs one OS thread for every bound thread; additionally, there is a variable number of so-called "worker" OS threads that are used to execute the unbounded (lightweight) threads." How does the runtime system determine the number of worker threads? Is the number hardcoded in the RTS or dynamically adjustable? Can a programmer specify it as an RTS option or change it using an API? I would like to use a large number (say, 2000) of unbounded threads, each calling a blocking, safe foreign function via FFI import. What is supposed to happen if all the worker threads are used up? I tried this in the recent GHC 6.5 and got some kind of "runaway worker threads?" RTS failure message when more than 32 threads are used. Is it a current limitation of the RTS, or should I file a bug report for it? Thanks, Peng

8 22

partial x86_64 binary snapshots?
by Jens Petersen 22 Jun '06

22 Jun '06

I wanted to download a nightly binary stable snapshot for Linux x86_64 to test if the problem with selinux on Fedora Core has been fixed... but AFAICT recent snapshots (I tried too) are broken: % bunzip2 -t ghc-6.4.2.20060621-x86_64-unknown-linux.tar.bz2 bunzip2: ghc-6.4.2.20060621-x86_64-unknown-linux.tar.bz2: file ends unexpectedly You can use the `bzip2recover' program to attempt to recover data from undamaged sections of corrupted files. Just wondering what is up with this? Jens

1 1

RE: about threads (here: increasing thread IDs)
by Jost Berthold 21 Jun '06

21 Jun '06

Hi Seth, perhaps this will help you, some information from the implementation side concerning your question about threadIDs: Related code lived in rts/Schedule.c, has recently moved to rts/Threads.* in the head. The ThreadID assigned by the RTS is continuously increasing, and declared to be StgInt32 (32 bits usually). So you might get int overflows when running your server for a *fairly* long time. OTOH, depending on what you want to do with the ThreadIDs that forkIO returns, IMHO you could just *ignore* these overflows. From your short description of how the servers work, you know an upper bound on the no. of running threads (6), but not on their time to live. So IMHO the worst case scenario would be: a new thread for a client gets ID n ... 2^32 other server threads get spawned another new thread for a client gets the same ID n again ... one of the threads supposed to terminate (killThread n) AFAICT, the killThread action will arbitrarily choose one of the two. HTH Jost ------ you wrote ------ Date: Tue, 20 Jun 2006 23:22:26 -0400 From: Seth Kurtzberg <seth(a)cql.com> Subject: Re: FFI: number of worker threads? To: "Li, Peng" <ringer9cs%2Bghc(a)gmail.com> Cc: glasgow-haskell-users(a)haskell.org Another related question. I have some threaded applications running which are servers and run continuously. A thread is spawned for each new connection, and the thread exits when the client terminates. I've noticed that the thread ID increases. On one process I checked today I am up to thread number 3300. The number of running threads is not increasing; only six threads are running on this particular process. The threads are cleaned up and exit. The thread _ID_ is continually increasing. Is this going to cause a problem when the thread ID exceeds some value? Do I have to force the server process to recycle periodically? ...

2 1

The GHC hackathon: 14,15 September 2006, Portland
by Simon Peyton-Jones 21 Jun '06

21 Jun '06

Friends After some encouragement at last year's ICFP, Simon and I are going to run a GHC hackathon, in Portland, just before ICFP this September (14-15th). It'll be held at Galois's offices, in Beaverton; we're very grateful to Galois for hosting the meeting. The details are here: http://hackage.haskell.org/trac/ghc/wiki/Hackathon If you are interested in finding out a bit about how GHC works inside, then you should find the hackathon fun. It will certainly be informal and interactive. GHC is a 100% open-source project, and the more people that are involved the better it goes. If you think you might come, please take a look at the above page, and let us know by registering. See you soon! Simon

1 0

Would you like a job working on GHC?
by Simon Peyton-Jones 21 Jun '06

21 Jun '06

Would you like to be paid to work on GHC? The Glasgow Haskell Compiler (GHC) is now being used by so many people, on so many platforms, that at GHC HQ we've been struggling to keep up. We're delighted to say that we are now looking to hire a GHC support engineer to work with us. Here is how we envisage working together: * You would work 3-5 days a week on GHC (details flexible). * You don't need to live in Cambridge, but we would like you to visit every few months, so that we build up a face-to-face relationship. * You'll be a self-employed consultant, not a Microsoft employee. You bill us; we pay you. * The position is a one year fixed term contract in the first instance. * GHC is, and will remain, an open-source project with a BSD-style license. We are looking for help with the role of supporting GHC in the field. Specifically, here's what the job involves: * Investigating, prioritising and fixing bugs, improving the test suite * Managing the STABLE branch of the source tree * Managing the GHC release cycle * Keeping the GHC web site up to date * Answering questions from GHC users * Helping to support the open-source community that works on GHC itself We're particularly looking for someone who is experienced in a variety of operating system platforms and libraries. You should be able to deal with questions like "I try to build GHC on Solaris 2.3.4 and get `undefined symbol _readline'"; or "How do I link to DLLs on Windows?"; or "does the ByteString library in GHC 6.2.1 do XXX?". The role will not primarily involve working on GHC's core; it's the parts round the edges that you would mainly focus on. You must be able to read and write Haskell, but we don't expect that you'll need to write a great deal of Haskell code. C and 'make' experience are important too. Above all, we are looking for someone who is enthusiastic about Haskell, and fired up about the prospect of becoming a GHC expert. If you are interested, please send your CV and a statement of why you would be a good fit for the job, to our Human Resources Department at camjobop(a)microsoft.com We're happy to accept informal enquiries of course: please contact Simon Marlow (simonmar(a)microsoft.com) or Simon Peyton Jones (simonpj(a)microsoft.com) for further information. The rate of pay is dependent on qualifications and experience. It'll be fun! Simon Peyton Jones and Simon Marlow

1 1

Left-to-right bias in type checking GADT functions
by Stefan Holdermans 20 Jun '06

20 Jun '06

Hi all, I have to admit that I haven't checked if this is a known issue... The following is tested with both GHC 6.4.1 and 6.4.2; I have not checked the HEAD. Let me define a simple GADT |F|, -- a simple GADT data F :: * -> * where F :: Int -> F () and a more or less trivial function |g| operating on it: -- type checks g :: F a -> a -> Int g (F n) () = n No problems up to here. But then let me just flip the parameters of | g|---and call the resulting function |h|: -- does not type check h :: a -> F a -> Int h () (F n) = n -- !!! Now, this definition of h does not type check: Couldn't match the rigid variable `a' against `()' `a' is bound by the type signature for `h' Expected type: a Inferred type: () When checking the pattern: () In the definition of `h': h () (F n) = n So, it seems that there's a left-to-right bias in type checking functions over GADTs. I cannot imagine this being by design, for it certainly seems possible to type check functions of this sort in a conceptually unbiased fashion. Anyone to shine a light on this? TIA, Stefan ------------------------------------------ Bias.lhs: > {-# OPTIONS -fglasgow-exts #-} > -- a simple GADT > data F :: * -> * where F :: Int -> F () > -- type checks > g :: F a -> a -> Int > g (F n) () = n > -- does not type check > h :: a -> F a -> Int > h () (F n) = n -- !!!

2 2

[glasgow-haskell-users] Searchpath vs. Cabalfind (was: installing a Cabal package non-exposed)
by Dimitry Golubovsky 17 Jun '06

17 Jun '06

Alex, S. Alexander Jacobson wrote: > Searchpath allows you easily to give each application its own > namespace and automates import chasing accross the internet. It is a > work in progress and is definitely less refined than cabal, but if you > are having this sort of issue you might want to try it out. See > http://searchpath.org Looks very interesting. However I couldn't get anything like "https://hackage.org/modules.map" in a browser (nor with http: nor from www.hackage.org) it says "unknown host". I am just curious: instead of querying a centralized database, have you tried any way to get modules' locations from search engines? I wish I could see what "modules.map" looks like, then I could see better whether what Cabalfind gets from Google could be an alternative? Not that I am pushing for Cabalfind in any way; just curious. -- Dimitry Golubovsky Anywhere on the Web

2 1