February 2002 - Glasgow-haskell-users

RE: GHC on Win32 strangeness
by Simon Peyton-Jones 11 Feb '02

11 Feb '02

I've updated the instructions on win32 building, but they havn't got to the web site yet. You have to say ./configure --host=i386-unknown-mingw32 --with-gcc=/mingw/bin/gcc Without this, things go badly arwy. For the --with-gcc part you need to install a mingwin bundle from http://www.mingw.org Simon | -----Original Message----- | From: Manuel M. T. Chakravarty [mailto:chak@cse.unsw.edu.au] | Sent: 11 February 2002 03:24 | To: glasgow-haskell-users(a)haskell.org | Cc: brk(a)jenkon.com | Subject: GHC on Win32 strangeness | | | Hi! | | Bryn <brk(a)jenkon.com> is trying to compile C->Haskell with | GHC on Win32, but ran into some problems. Firstly, | configure says, | | $ ./configure | loading cache ./config.cache | checking host system type... i686-pc-cygwin | checking target system type... i686-pc-cygwin | checking build system type... i686-pc-cygwin | checking for Cygwin environment... (cached) no | checking for gcc... (cached) gcc | [..] | | So, it says no Cygwin, but identifies the system as | i686-pc-cygwin. AFAIK, GHC uses mingw32 these days, doesn't | it? How can this be reliably tested for? (To exclude | -package posix on that platform.) | | Secondly, are there any issues with symlinks? I append a | sequence of commands that seems to indicate that GHC doesn't | find a file, because it is symlinked. | | Any ideas? | | Cheers, | Manuel | | -=- | | Administrator@BRK ~/hipar/build/ghc5/base/admin | $ file BaseVersion.hs | BaseVersion.hs: symbolic link to ../../../../base/admin/BaseVersion.hs | | So, change it to an absolute: | | Administrator@BRK ~/hipar/build/ghc5/base/admin | $ ln -s /home/hipar/base/admin/BaseVersion.hs BaseVersion.hs | | Administrator@BRK ~/hipar/build/ghc5/base/admin | $ ls | BaseVersion.hs Common.hs Config.hs Makefile | | Administrator@BRK ~/hipar/build/ghc5/base/admin | $ file BaseVersion.hs | BaseVersion.hs: symbolic link to /home/hipar/base/admin/BaseVersion.hs | | The build fails the same way, copy the command line and | restrict it to just the file we changed, to be sure: | | Administrator@BRK ~/hipar/build/ghc5/base | $ ghc -M -syslib lang -optdep-f -optdep.depend -iadmin | -ierrors -igeneral -igra phs -istate -isyms -isyntax -isysdep | admin/BaseVersion.hs | ghc.exe: file `admin/BaseVersion.hs' does not exist | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-| haskell-users |

1 0

GHC on Win32 strangeness
by Manuel M. T. Chakravarty 10 Feb '02

10 Feb '02

Hi! Bryn <brk(a)jenkon.com> is trying to compile C->Haskell with GHC on Win32, but ran into some problems. Firstly, configure says, $ ./configure loading cache ./config.cache checking host system type... i686-pc-cygwin checking target system type... i686-pc-cygwin checking build system type... i686-pc-cygwin checking for Cygwin environment... (cached) no checking for gcc... (cached) gcc [..] So, it says no Cygwin, but identifies the system as i686-pc-cygwin. AFAIK, GHC uses mingw32 these days, doesn't it? How can this be reliably tested for? (To exclude -package posix on that platform.) Secondly, are there any issues with symlinks? I append a sequence of commands that seems to indicate that GHC doesn't find a file, because it is symlinked. Any ideas? Cheers, Manuel -=- Administrator@BRK ~/hipar/build/ghc5/base/admin $ file BaseVersion.hs BaseVersion.hs: symbolic link to ../../../../base/admin/BaseVersion.hs So, change it to an absolute: Administrator@BRK ~/hipar/build/ghc5/base/admin $ ln -s /home/hipar/base/admin/BaseVersion.hs BaseVersion.hs Administrator@BRK ~/hipar/build/ghc5/base/admin $ ls BaseVersion.hs Common.hs Config.hs Makefile Administrator@BRK ~/hipar/build/ghc5/base/admin $ file BaseVersion.hs BaseVersion.hs: symbolic link to /home/hipar/base/admin/BaseVersion.hs The build fails the same way, copy the command line and restrict it to just the file we changed, to be sure: Administrator@BRK ~/hipar/build/ghc5/base $ ghc -M -syslib lang -optdep-f -optdep.depend -iadmin -ierrors -igeneral -igra phs -istate -isyms -isyntax -isysdep admin/BaseVersion.hs ghc.exe: file `admin/BaseVersion.hs' does not exist

2 1

Re: efficiency question
by Jorge Adriano 08 Feb '02

08 Feb '02

> I'd say that's because in the second case you also got to apply the (,), > besides the (+)/(-) constructor during the transversing... > Am I right? opss... I meant to write: the (,) constructor besides the (+)/(-)... J.A.

1 0

efficiency question
by Hal Daume III 08 Feb '02

08 Feb '02

define test1 l = let s1 = foldr (+) 1 l s2 = foldr (-) 1 l in (s1, s2) test2 l = let s = foldr (\x (a,b) -> (x+a,x-b)) (1,1) l in s why is test1 so much faster than test2 for long lists l (eg [1..1000000])? replacing foldr with foldl makes it faster (of course), but test2 is still much slower. i *expected* test2 to be much faster because you're only traversing the list once. presumably the two elements "a" and "b" in test2 could be put in registers and i'd imagine test2 should be faster (it certainly would be if written in c). - 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

ANNOUNCE: GHC 5.03.20020204 snapshot release
by Julian Seward (Intl Vendor) 08 Feb '02

08 Feb '02

There have been a number of significant improvements made to GHC since the 5.02 sources were branched off the main GHC development tree, but we're not quite ready to make a full 5.04 release yet. However, we're keen to get feedback on the new features from people that don't have easy access to CVS or don't have the time & patience to build GHC from scratch, so we're making a snapshot release of the current GHC, in source and binary form. There are NO GUARANTEES as to the stability of this release, although we do know it has passed our basic three-stage bootstrap test and run the regression tests successfully. In some cases the documentation hasn't been fully updated to reflect the new features yet. Briefly, the changes relative to 5.02.2 are: - The type system now supports arbitrary rank polymorphism, given appropriate type annotations. - Heap profiling has had a major overhaul and now supports retainer profiling and biographical profiling ala nhc98. - Major improvements to the native code generators. You can now compile any and all code through them, including the Prelude. - The FFI syntax has been updated to match the latest version of the FFI Haskell 98 Addendum. - newtypes support deriving *any* class for which the underlying type is also an instance. - Linear implicit parameters: a highly experimental feature. - GHCi has several new commands: - ':show bindings' to list the bindings made on the command line, - ':show modules' to show which modules are loaded, - ':module' has been enhanced as per the discussion on the mailing list (syntax is still experimental - feedback welcome). - ':browse' similar to Hugs' :browse command. And many other minor changes & bugfixes. At the moment there is just the sources and a binary build for x86-linux available. We may add builds for other if they become available. You can get it from http://haskell.cs.yale.edu/ghc/download_ghc_snapshot.html Enjoy! The GHC Development Team.

2 1

RE: inscrutable warning
by Simon Peyton-Jones 08 Feb '02

08 Feb '02

It's because there's an application of a strict function, the dictionary selector 'a'. In general, if f is strict then f (g x) is turned into case g x of y { DEFAULT -> f y } But it is rather inscrutable. It didn't arise from a strictness annotation put in by the user. The right thing to do is to fix the bytecode generator. Simon | -----Original Message----- | From: Julian Seward (Intl Vendor) | Sent: 08 February 2002 13:55 | To: 'Dean Herington'; glasgow-haskell-users(a)haskell.org | Cc: Simon Peyton-Jones | Subject: RE: inscrutable warning | | | | The warning says that the interpreter has ignored a | polymorphic case, since it's hard to implement. Your program | should still behave the same, although possibly less strictly | than you intended. | | What really mystifies us is why there is such a thing in the | program in the first place. Using -ddump-sat we can see that | there is indeed a | case ... of { DEFAULT -> ... } | but how it came to be, I do not know. | | We may investigate further. | | J | | | Can someone explain the following warning? | | | | ------------------------------- | | module Warning where | | | | type A = IO () | | | | class CA t | | where a :: t -> A | | instance CA () | | where a = return | | instance (CA t) => CA (IO t) | | where a = (>>= a) | | | | data B = B A | | | | class CB t | | where b :: t -> B | | instance (CA t) => CB (IO t) | | where b = B . a | | ------------------------------- | | | | % ghci Warning.hs | | ___ ___ _ | | / _ \ /\ /\/ __(_) | | / /_\// /_/ / / | | GHC Interactive, version 5.02.2, | | for Haskell | | 98. | | / /_\\/ __ / /___| | http://www.haskell.org/ghc/ | | \____/\/ /_/\____/|_| Type :? for help. | | | | Loading package std ... linking ... done. | | Compiling Warning ( Warning.hs, interpreted ) | | WARNING: ignoring polymorphic case in interpreted mode. | | Possibly due to strict polymorphic/functional constructor args. | | Your program may leak space unexpectedly. | | | | Ok, modules loaded: Warning. | | Warning> | | | | _______________________________________________ | | Glasgow-haskell-users mailing list | | Glasgow-haskell-users(a)haskell.org | | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users | | |

1 0

RE: inscrutable warning
by Julian Seward (Intl Vendor) 08 Feb '02

08 Feb '02

The warning says that the interpreter has ignored a polymorphic case, since it's hard to implement. Your program should still behave the same, although possibly less strictly than you intended. What really mystifies us is why there is such a thing in the program in the first place. Using -ddump-sat we can see that there is indeed a case ... of { DEFAULT -> ... } but how it came to be, I do not know. We may investigate further. J | Can someone explain the following warning? | | ------------------------------- | module Warning where | | type A = IO () | | class CA t | where a :: t -> A | instance CA () | where a = return | instance (CA t) => CA (IO t) | where a = (>>= a) | | data B = B A | | class CB t | where b :: t -> B | instance (CA t) => CB (IO t) | where b = B . a | ------------------------------- | | % ghci Warning.hs | ___ ___ _ | / _ \ /\ /\/ __(_) | / /_\// /_/ / / | | GHC Interactive, version 5.02.2, | for Haskell | 98. | / /_\\/ __ / /___| | http://www.haskell.org/ghc/ | \____/\/ /_/\____/|_| Type :? for help. | | Loading package std ... linking ... done. | Compiling Warning ( Warning.hs, interpreted ) | WARNING: ignoring polymorphic case in interpreted mode. | Possibly due to strict polymorphic/functional constructor args. | Your program may leak space unexpectedly. | | Ok, modules loaded: Warning. | Warning> | | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users |

1 0

Unsure about 'IO'
by Alex van Ballegooij 06 Feb '02

06 Feb '02

Dear all, We've been working on bulding an interpreter in haskell and we seem to have encountered a problem. At the end of the email I've pasted a minimal version of the program, that shows the error. The problem is that we 'putStr' a prompt onto (unbuffered) stdout and then read a line, process it and 'print' output, all very basic. This is done in the function 'processline' and works fine is this function is used directly from 'main', however, if used from within the 'while' construct it fails. In the 'while' context it appears that the 'getLine' is executed before the 'putStr' is (or perhaps stdout is not as 'unbuffered' as we expect ?). As all the functions in this example are sequenced in the context of the IO monad, I guess that order of execution should be predefined, right ? Any help would be greatly appreciated, just for information: we tested with GHC 5.02.1 and 5.02.2. Kind regards, Alex... ---- import IO import Monad notM :: Monad m => m Bool -> m Bool notM = liftM not processline :: IO () processline = do putStr "prompt: " line <- getLine putStrLn ("Read: " ++ line) while :: Monad m => m Bool -> m a -> m () while t s = do b <- t if(b) then do { s; while t s; return () } else do { return () } main :: IO () main = do hSetBuffering stdout NoBuffering putStrLn "Single call:" processline putStrLn "Call from 'while' construct:" while (notM isEOF) (processline)

3 2

RE: Unsure about 'IO'
by Simon Marlow 06 Feb '02

06 Feb '02

> We've been working on bulding an interpreter in haskell > and we seem to have encountered a problem. > At the end of the email I've pasted a minimal version of the program, > that shows the error. > > The problem is that we 'putStr' a prompt onto (unbuffered) stdout > and then read a line, process it and 'print' output, all very basic. > > This is done in the function 'processline' and works fine is this > function is used directly from 'main', however, if used from within > the 'while' construct it fails. > In the 'while' context it appears that the 'getLine' is executed > before the 'putStr' is (or perhaps stdout is not as 'unbuffered' > as we expect ?). In your example it is the 'isEOF' that is waiting for input. The I/O system can't determine whether the end of the input has been reached yet without waiting for the user to type something: the user might type ^D which would indicate EOF, or they might type some characters. This is arguably wrong, although it does seem to follow the Haskell Library Report (section 11.4.1). Strictly speaking, it would be wrong not to block in isEOF because we might be at the end of file and not know it yet. You can use hReady to determine whether isEOF will block (although interestingly the report says that hReady can fail with an EOF error; GHC won't do this in practice). Cheers, Simon

1 0

inscrutable warning
by Dean Herington 06 Feb '02

06 Feb '02

Can someone explain the following warning? ------------------------------- module Warning where type A = IO () class CA t where a :: t -> A instance CA () where a = return instance (CA t) => CA (IO t) where a = (>>= a) data B = B A class CB t where b :: t -> B instance (CA t) => CB (IO t) where b = B . a ------------------------------- % ghci Warning.hs ___ ___ _ / _ \ /\ /\/ __(_) / /_\// /_/ / / | | GHC Interactive, version 5.02.2, for Haskell 98. / /_\\/ __ / /___| | http://www.haskell.org/ghc/ \____/\/ /_/\____/|_| Type :? for help. Loading package std ... linking ... done. Compiling Warning ( Warning.hs, interpreted ) WARNING: ignoring polymorphic case in interpreted mode. Possibly due to strict polymorphic/functional constructor args. Your program may leak space unexpectedly. Ok, modules loaded: Warning. Warning>

1 0