August 2006 - Glasgow-haskell-users

Default bufferring behavior
by Vyacheslav Akhmechet 23 Aug '06

23 Aug '06

I encountered the following issue last night. Writing a program like this: main = do putStr "> " x <- getLine putStr x behaves as expected in hugs and ghci (first prints "> ", then waits for input, then prints the input). However, when compiling and running an executable via ghc, the behavior was very different. I saw a blank console screen and when I typed something it was echoed back to me (as in "> 25") only after I hit enter. This is trivially fixed by flushing: main = do putStr "> " hFlush stdout x <- getLine putStr x or by turning off buffering. Is this the intended behavior? This problem does not appear when using putStrLn (as it flushes automatically) and is quite distracting to newbies. To me it looks like this behavior is at odds with something one can reasonably expect and bufferring should be turned off by default. What do you guys think? Thanks, - Slava.

2 1

Re[2]: [GHC] #814: RTS always grabs 256Mb on startup
by Bulat Ziganshin 23 Aug '06

23 Aug '06

Hello GHC, Wednesday, August 23, 2006, 2:21:30 AM, you wrote: > #814: RTS always grabs 256Mb on startup > * resolution: => fixed > * status: new => closed > Fixed for 6.6 Esa, i thanks you for fixing this problem, it was really uncomfortable for me. can you please also apply your patch to 6.4 sources? -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

1 0

Circular dependency between the compiler and TH?
by Joel Reymont 22 Aug '06

22 Aug '06

Apparently, pkString is part of TH.Syntax which is part of ./ libraries/template-haskell/Language/Haskell/TH/Syntax.hs. This does not get built after the compiler is built but the compiler depends on it in compiler/hsSyn/Convert.lhs. How do I resolve this? -- http://wagerlabs.com/

6 18

Template Haskell crashes unexpectedly...
by Brian Hulley 22 Aug '06

22 Aug '06

Hi, I'm starting to explore Template Haskell and I've got the following program: -- almost directly from "Template meta-programming in Haskell" paper module Duma.Template.Test ( Format(..) , printf ) where import Language.Haskell.TH data Format = D | S | L String printf :: [Format] -> ExpQ printf fs = gen fs [| "" |] gen :: [Format] -> ExpQ -> ExpQ gen [] x = x gen (D : xs) x = [| \n -> $(gen xs [| $x ++ show n |]) |] gen (S : xs) x = [| \s -> $(gen xs [| $x ++ s |] ) |] gen (L s : xs) x = gen xs [| $x ++ $(stringE s) |] module Main where import Duma.Template.Test import Language.Haskell.TH data Car = Car {_f :: Int} getInfo :: Q Info getInfo = reify (mkName "Car") main = do e <- runQ (printf [D,S,L "foo"]) putStrLn (pprint e) let x = $(printf [D,S,L "foo"]) 10 "hello" putStrLn x -- Crashes if I try to print out the info -- info <- runQ getInfo -- putStrLn (pprint info) _ <- getChar return () The example from the paper works fine with the few minor adjustements ie Expr --> ExpQ, lift --> stringE. However if I try and obtain the Info it compiles but then crashes at runtime (using ghc 6.4.2 on Windows). Any ideas? (Perhaps because the type doesn't exist - yet the first call to runQ (printf...) *does* work at runtime) Also, I'm puzzled by the type of reify, because the name "Car" above should surely be both a TyConI and a DataConI so how does this function decide which to return the info about? I expected an extra parameter to determine what namespace to look the name up in. BTW what I'm trying to do is write a TH function which adds a field to a data type and a corresponding instance method eg given: class Object a where getName :: a -> Unique foo :: a -> Int data Square = Square {_f :: Int} instance Object Square where foo = _f will modify the above code (or create new instance and new data type) to: data Square = Square {_name :: Unique, _f :: Int} instance Object Square where getName = _name foo = _f ie adding a bit of boilerplate to the data type and instance decl. Anyway any ideas about why reify doesn't distinguish between lookups of name-as-tycon vs name-as-datacon or why it crashes will be greatly appreciated ;-) Thanks, Brian. -- Logic empowers us and Love gives us purpose. Yet still phantoms restless for eras long past, congealed in the present in unthought forms, strive mightily unseen to destroy us. http://www.metamilk.com

3 7

[GHC Users] GLUT & OpenGL on Intel Mac
by Ryan Cleven 22 Aug '06

22 Aug '06

I was wondering if anyone has managed to get the GLUT or OpenGL trivial examples going on the Intel mac binary distribution from aug 8th running? I can't get them running under either ghc or ghci. Thanks, Ryan

1 0

GHCi hacking
by Ulf Norell 21 Aug '06

21 Aug '06

Hi. I want to use GHCi as the interface to my Application. The simple solution is to have the Application store its state in global IORefs. A user can then start up ghci with -package Application and use the provided functions to communicate with the Application. This works nicely. Now the tricky part. I want the user to be able to implement her own layers on top of the Application API. A typical user interaction could look something like this: $ ghci -package Application *Application:API> startApplication *Application:API> :load UserLayer *main:UserLayer> myCleverFunction 42 The problem, of course, is that as soon as the user says ':load' ghci forgets all about the state of the Application. My solution to the problem is to compile my own version of ghci (copy InteractiveUI.hs and use -package ghc) and remove the call to rts_revertCAFs when loading new modules. This seems to work, but since I don't really have a clue what I'm doing I wanted to ask a few questions: 1) What is a CAF? 2) What breaks down if you don't revert them? 3) Does not reverting the CAFs really solve my problem? / Ulf

5 6

Mac/PPC threaded RTS problems -- potential clue
by David Kirkman 21 Aug '06

21 Aug '06

I managed to build ghc-6.5.20060804 on a powerpc mac, and I spent a little time Saturday night trying to work out what might be going on with the threaded RTS. Running the testsuite with make WAY=threaded1 shows many (73) failures. Many of them (the conc??? set) have to do with TVars, in particular writing TVars. On my dual-proc G5 macs, with ghc-6.5.20060804, the following program will hang when running with the threaded RTS, but will work fine when compiled without -threaded. >module Main where > >import GHC.Conc >import Control.Concurrent > >main = do > t1 <- atomically (do t1 <- newTVar 80 > return t1) > atomically ( do writeTVar t1 1 ) > putStr ("done\n") This problem seems to be powerpc specific -- it works fine with the threaded RTS on a multi-processor intel mac (built by me, from the same ghc-6.5.20060804 source tree I used to build the ppc compiler). Looking around in STM.c (via decidedly low-tech printfs) I quickly zoomed in on cond_lock_tvar, which lead me to cas (atomic compare and swap) in SMP.h, where I found (I think) a fairly clear error in the powerpc code -- I've appended a patch to the end of this message. The problem is that the inline assembler code was placing the result of an operation in a register that is used as input later in the code. At the bottom of this message I've extracted a short short code fragment that you can run through gcc (on a powerpc machine) to see the generated assembly output. The changes to fix the problem are fairly simple. The first adds an ampersand to the output list of the assembly fragment ("=r" (result) --> "=&r" (result)) The ampersand just tells gcc that result can not be placed in a register used for any of the input parameters (o, n, or p). Otherwise, it feels free to place output parameters in the same registers used by the inputs -- but because of the flow of control here we need everything in a distinct register. This change fixes the TVar program above. The second change adds a clobber list (the :"cc", "memory"). This tells gcc that the condition code (due to the compare) and memory (due to the store) might be changed during the asm execution. The lack of a clobber list did not seem to be causing any trouble, but without it gcc is free to assume that no state is changed during the execution. Applying the following patch to SMP.h, and rebuilding everything, I not only fixed the simple writeTVar program, but it also fixed 8 programs in the testsuite (conc043 -> conc049, conc052 and conc053). The only conc test program that still fails is conc039. But there are still many mac problems, I still have 132 unexpected failures with make fast. At least the patch does not cause any new failures (in either make fast or make WAY=threaded1). Anyway, seeing as the change to SMP.h fixes a fair number of test cases in the testsuite, I figure there is some chance that it might fix some of the problems that people are having with the threaded RTS. On the other hand, I'm not real happy the the large number of testsuite failures my build gets, so I can't really call this 'tested'. But I'm posting because it might be a useful clue for somebody with a little more mac/ghc experience. Cheers, -david k. // // Short code to run through gcc -S. On my powermac, without the change // the generated assembly produces // 1: lwarx r0, 0, r0 // load to here ---^ ^------- from address here // But we need this value for the stwcx. // // with the fix, the first line of the generated assembly becomes // 1: lwarx r11, 0, r0 // and r0 remains unmodified if we need to use it later in the stwcx. /* * CMPXCHG - the single-word atomic compare-and-exchange instruction. Used * in the STM implementation. */ long cas(long* p, long o, long n) { long result; // // Change // :"=r" (result) --> :"=&r" (result) // to get result and p in different registers __asm__ __volatile__ ( "1: lwarx %0, 0, %3\n" " cmpw %0, %1\n" " bne 2f\n" " stwcx. %2, 0, %3\n" " bne- 1b\n" "2:" :"=r" (result) :"r" (o), "r" (n), "r" (p) ); return result; } Here's a "diff -cp" for SMP.h *** SMP.h Sun Aug 13 01:08:53 2006 --- SMP-new.h Sun Aug 13 01:08:47 2006 *************** cas(StgVolatilePtr p, StgWord o, StgWord *** 76,83 **** " stwcx. %2, 0, %3\n" " bne- 1b\n" "2:" ! :"=r" (result) :"r" (o), "r" (n), "r" (p) ); return result; #else --- 76,84 ---- " stwcx. %2, 0, %3\n" " bne- 1b\n" "2:" ! :"=&r" (result) :"r" (o), "r" (n), "r" (p) + :"cc", "memory" ); return result; #else

3 2

Latest ghc build error
by Joel Reymont 21 Aug '06

21 Aug '06

/usr/local/bin/ghc -H16m -O -istage1/utils -istage1/basicTypes - istage1/types -istage1/hsSyn -istage1/prelude -istage1/rename - istage1/typecheck -istage1/deSugar -istage1/coreSyn -istage1/ specialise -istage1/simplCore -istage1/stranal -istage1/stgSyn - istage1/simplStg -istage1/codeGen -istage1/main -istage1/ profiling -istage1/parser -istage1/cprAnalysis -istage1/ ndpFlatten -istage1/iface -istage1/cmm -istage1/nativeGen - istage1/ghci -Istage1 -DGHCI -DBREAKPOINT -package template-haskell - threaded -package readline -DUSE_READLINE -cpp -fglasgow-exts -fno- generics -Rghc-timing -I. -IcodeGen -InativeGen -Iparser -package unix -ignore-package lang -recomp -Rghc-timing -H16M '-#include "cutils.h"' -i../compat -ignore-package Cabal -fno-cse -c ghci/ InteractiveUI.hs -o stage1/ghci/InteractiveUI.o -ohi stage1/ghci/ InteractiveUI.hi /tmp/ghc17177_0.hc: In function 'smEl_ret': /tmp/ghc17177_0.hc:42472:0: error: void value not ignored as it ought to be /tmp/ghc17177_0.hc: In function 'smEf_ret': /tmp/ghc17177_0.hc:42504:0: error: void value not ignored as it ought to be <<ghc: 812568504 bytes, 1358 GCs, 15125404/31986692 avg/max bytes residency (10 samples), 80M in use, 0.00 INIT (0.00 elapsed), 2.39 MUT (10.94 elapsed), 1.75 GC (1.86 elapsed) :ghc>> -- http://wagerlabs.com/

2 1

Instruction to build current GHC 6.5 on Mac OS X/Intel x86
by Esa Ilari Vuokko 21 Aug '06

21 Aug '06

Hi, Paolo Martini got ghc to built using Audrey Tang's build from 20060608 on Mac OS X/Intel x86. Instructions are available at ghc trac: http://hackage.haskell.org/trac/ghc/wiki/X86OSXGhc Please edit the page if you have corrections. Best regards, --Esa Ilari Vuokko

2 1

Re: [Haskell] Haskell Workshop 2006 Call for participation
by Bulat Ziganshin 21 Aug '06

21 Aug '06

Monday, August 14, 2006, 10:34:22 AM, Andres Loeh wrote: > 14:30 Simon Marlow (Microsoft Research) > An Extensible Dynamically-Typed Hierarchy of Exceptions is this planned to be included in ghc 6.6? 6.8? > 15:00 David Himmelstrup (Denmark) > Demo: Interactive Debugging with GHCi and this? -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

2 1