October 2006 - Glasgow-haskell-users

recursive commentaries :)
by Bulat Ziganshin 19 Oct '06

19 Oct '06

Hello glasgow-haskell-users, http://hackage.haskell.org/trac/ghc/wiki/Commentary contains a link 'The old GHC Commentary' which points to the http://www.cse.unsw.edu.au/~chak/haskell/ghc/comm/ page. guess what this page contains? :) it will be great to restore link to old commentaries -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

1 0

Re: Bad control characters in GHCi 6.6 (Mac OS X PPC)
by Björn Buckwalter 19 Oct '06

19 Oct '06

On 10/16/06, Simon Marlow <simonmarhaskell(a)gmail.com> wrote: > Björn Buckwalter wrote: > > > I downloaded the GHC 6.6 tarballs, built and installed them on my mac > > (Tiger, PPC) using the documented procedure: > > ./configure > > make > > make install > > > > The build and install was fairly smooth. However, ghci will not > > recognize "control characters", e.g. delete/backspace, arrow keys, > > ctrl-a, ctrl-d et c. It seems to recognize ctrl-c for cancellin5.2g a > > line but will print the ^C first. A typical session may look like: > > It sound like either you don't have readline installed, or the GHC build didn't > detect it. Do you have readline? (the real GNU readline, not Apple's version). I have GNU readline 5.1 installed through darwinports. As for the GHC build detecting it I wouldn't know. Rerunning ./configure produces the following lines mentioning readline (note that I have already built and installed GHC 6.6 and have it in the path): checking whether ghc has readline package... no checking readline/readline.h usability... yes checking readline/readline.h presence... yes checking for readline/readline.h... yes checking readline/history.h usability... yes checking readline/history.h presence... yes checking for readline/history.h... yes configure: configuring in readline checking for readline in -lreadline... yes checking for rl_readline_version in -lreadline... no config.status: creating readline.buildinfo config.status: creating include/HsReadlineConfig.h config.status: include/HsReadlineConfig.h is unchanged I don't know what to make of the above. Thanks, Bjorn Buckwalter

1 0

Data.Array.MArray "lost" bounds
by Andreas Marth 18 Oct '06

18 Oct '06

With ghc-6.6 there is a change in module Data.Array.MArray which now includes a function getBounds :: (MArray a e m, Ix i) => a i e -> m (i,i) but it misses the good old bounds [which had type bounds :: (HasBounds a, Ix i) => aie -> (i,i)]. What is the rational behind this change? Are MArrays now able to change there size? Thanks for any eplanations, Andreas

3 2

Some potentially useful installers..
by Sigbjorn Finne 17 Oct '06

17 Oct '06

Hi, I've bundled up a bunch of Win32 installers for various tools that come in handy when developing with&for GHC: Alex, Happy, and Haddock (aka "The Marlow Collection") http://galois.com/~sof/msi/alex-2-0-1.msi http://galois.com/~sof/msi/happy-1-15.msi http://galois.com/~sof/msi/haddock-0-7.msi enjoy, --sigbjorn

2 2

typo in http://www.haskell.org/onlinereport/prelude-index.html
by Christian Maeder 17 Oct '06

17 Oct '06

I've just noticed the wrong example entry for break: break :: (a -> Bool) -> [a] -> ([a], [a]) break (<2) [1,2,3] = ([1],[2,3]) this should be changed to: break (>2) [1,2,3] = ([1,2],[3]) C. Prelude> break (<2) [1,2,3] ([],[1,2,3])

1 0

foldl laziness support. Reply
by Serge D. Mechveliani 17 Oct '06

17 Oct '06

Concerning the laziness support problem, I thank people for explanations about foldl and foldr. >> I wonder how to avoid these numerous cost pitfalls. >> Maybe, the complier could do more optimization? Duncan Coutts <duncan.coutts(a)worc.ox.ac.uk> writes > There are important differences between foldl, foldl' and foldr. It is > quite important to choose the right one. I don't think this can be done > automatically. > > In my experience, the choice is almost always between foldl' and foldr. > > [..] I do not see foldl' in the standard library. Is it of the GHC lib extension? has it strictness annotation? > So as Lemmih says, in this case you want to use foldr: > > import List (union) > main = let n = 10^4 :: Int > in > putStr > (shows (take 2 $ unionMany [[1 .. i] | i <- [1 .. n]]) "\n") > > unionMany = foldr union [] I see. Thank you. I have impression that something is here besides the intuition for the foldl/foldr choice. Here is a contrived example which is more close to my real situation. ----------------------------------------------------------------- import qualified Data.Set as Set (Set(..), empty, member, insert) import List (union, find) main = let n = 10^6 :: Int in putStr (shows (g1 n) "\n") f :: Int -> (Set.Set Int, [Int]) f n = -- original version, I write so because it is easy to program -- foldl add (Set.empty, []) [[1 .. i] | i <- [1 .. n]] where add (s, xs) ys = (Set.insert (sum xs) s, union xs ys) {- attempt to optimize (fails) -- h (Set.empty, []) [[1 .. i] | i <- [1 .. n]] where h (s, xs) [] = (s, xs) h (s, xs) (ys: yss) = h (Set.insert (sum xs) s, union xs ys) yss -} g1, g2 :: Int -> Bool -- client functions g1 n = case snd $ f n of x: _ -> even x _ -> False g2 n = let (set, xs) = f n in case find (> 100) xs of Just x -> Set.member (2*x) set _ -> False ----------------------------------------------------------------- Evidently, g1 n must have the cost of O(1). But in ghc-6.6 -O, it has O(n). How to improve f ? I tried foldr, and failed. The situation is so that some clients are as g1, and others are as g2, and, at least, g1 must be O(1). Regards, ----------------- Serge Mechveliani mechvel(a)botik.ru

4 3

scoped tyvar
by Bulat Ziganshin 17 Oct '06

17 Oct '06

Hello glasgow-haskell-users, should the following work in 6.6? -- | Alloc the mutable byte vector having `elems` elements of required type allocUnboxed :: (STorIO m s, Integral elems, Unboxed a) => elems -> m (MUVec s a) allocUnboxed elems :: m (MUVec s a) = -- it's line 149 allocUnboxedBytes (fromIntegral elems * sizeOfUnboxed (undefined::a)) the problem is that it works in 6.6rc1, but user reports that his compiler (i think, 6.6 release) can't compile it saying: Data/Unboxed.hs:149:0: Parse error in pattern may be it's because september's changes in scoped tyvars handling? how i can convert this code retaining compatibility with Hugs? -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

3 2

ghc for PC Solaris
by Christian Maeder 17 Oct '06

17 Oct '06

Hi, I would like to get ghc (version 6.6) working for our students. The machines are ATHLON-64 with Solaris 10. (SunOS 5.10 Generic_118855-19 i86pc i386 i86pc) If someone has a binary distribution of any ghc version (that is able to build ghc-6.6), please let me know. But I would also try to build ghc from hc-sources. What is the best platform I could get hc-sources from? We have sparc-sun-solaris2.10 and i686-pc-linux-gnu ourselves. In case that I have to port from another host platform as described in http://www.haskell.org/ghc/docs/6.6/html/building/sec-porting-ghc.html I've attached my generated target files from the "includes"-directory (with sources from the fresh ghc-6.6 release). Would someone create a "hc-file-bundle" on a more suitable host for me? Thanks for any help, Christian

2 2

Timing Stats in ghci-6.6
by Daniel Fischer 17 Oct '06

17 Oct '06

Hi, apparently the timing doesn't work properly in ghci 6.6 (at least for me), I always get very small timings, usually 0.01 secs, sometimes 0.00 secs, once 0.07 secs, no matter how long it actually took. Is that a common problem or is it only me? Cheers, Daniel

1 0

ghc-6.6 mac intel "binary bundle"
by Adam Megacz 17 Oct '06

17 Oct '06

Hi, I'm grateful to whoever prepared the binary bundle for ghc-6.6 on MacOS/Intel, but the instructions for installing it are missing/wrong. In particular, "make install" doesn't do much, and the only binary in the bundle called "ghc" appears to need additional configuration options (-B), so it seems to need some sort of installation procedure. The "INSTALL" file is not useful. Overall, I'm kind of uncertain exactly what this tarball contains. It looks like a "built" copy of the source code, but most of the source code is missing. It's not an installer (.mpkg) nor does it (apparently) contain any install scripts. Could you perhaps update this webpage with an example of the shell commands you might use to get it installed to the point where typing "ghc Foo.hs" works? http://www.haskell.org/ghc/download_ghc_66.html#macosxintel Thanks, - a -- PGP/GPG: 5C9F F366 C9CF 2145 E770 B1B8 EFB1 462D A146 C380

3 4