- Haskell-Cafe - Haskell.org

GHC 6.10, strange behaviour when profiling?
by Aleš Bizjak 12 Nov '08

12 Nov '08

Hello, I'm experiencing some strange behaviour with GHC 6.10 and would like an explanation why. Here's the problem. With GHC 6.8.[23] memory usage of a program was about 250mb (computing pi to 10^6 decimals). Now I tried recompiling and running with GHC 6.10 and I got more than 1.4gb before I killed it, so naturally I tried profiling but here's the strangeness. When I compile for profiling with -prof -auto-all and run with +RTS -p -RTS memory use is still about the same as before (> 1.4gb), but when I try running with +RTS -hc -RTS memory only reaches about 250mb. What is the explanation for this behaviour? I would send a program but it requires HERA, which is not on hackage ... Here is the output of -s ./Test 100000 +RTS -s 47,467,176 bytes allocated in the heap 18,046,776 bytes copied during GC 9,215,104 bytes maximum residency (6 sample(s)) 711,520 bytes maximum slop 80 MB total memory in use (11 MB lost due to fragmentation) Generation 0: 84 collections, 0 parallel, 0.01s, 0.01s elapsed Generation 1: 6 collections, 0 parallel, 0.01s, 0.01s elapsed INIT time 0.00s ( 0.00s elapsed) MUT time 3.66s ( 3.71s elapsed) GC time 0.02s ( 0.02s elapsed) RP time 0.00s ( 0.00s elapsed) PROF time 0.00s ( 0.00s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 3.68s ( 3.73s elapsed) %GC time 0.5% (0.6% elapsed) Alloc rate 12,982,553 bytes per MUT second Productivity 99.5% of total user, 97.9% of total elapsed ./Test 100000 +RTS -s -hc 47,471,160 bytes allocated in the heap 18,637,752 bytes copied during GC 13,580,144 bytes maximum residency (37 sample(s)) 810,712 bytes maximum slop 33 MB total memory in use (7 MB lost due to fragmentation) Generation 0: 85 collections, 0 parallel, 0.02s, 0.02s elapsed Generation 1: 37 collections, 0 parallel, 0.00s, 0.01s elapsed INIT time 0.00s ( 0.00s elapsed) MUT time 3.67s ( 3.73s elapsed) GC time 0.02s ( 0.03s elapsed) RP time 0.00s ( 0.00s elapsed) PROF time 0.00s ( 0.00s elapsed) EXIT time 0.00s ( 0.00s elapsed) Total time 3.70s ( 3.77s elapsed) %GC time 0.6% (0.8% elapsed) Alloc rate 12,941,165 bytes per MUT second Productivity 99.2% of total user, 97.4% of total elapsed -- Best regards, Aleš

3 4

Searching for ADT patterns with elem and find
by Paul Keir 12 Nov '08

12 Nov '08

Hi All, If I have an ADT, say data T = A String Integer | B Double | C deriving(Eq) and I want to find if a list (ts) of type T contains an element of subtype "B Double", must my "containsTypeX" function use a second "isTypeX" function as follows: isTypeB :: T -> Bool isTypeB (B _) = True isTypeB _ = False containsTypeB :: [T] -> Bool containsTypeB ts = maybe False (\x -> True) (find isTypeB ts) I understand that while something like "find C ts" will work, "find (isTypeB _) ts" will not, but is there no such thing as a pattern combinator(?), or lambda that could help with this situation. I find I have many individual "isTypeB" functions now. Regards, Paul

6 9

abstract extensible types?
by Alberto G. Corona 12 Nov '08

12 Nov '08

Is there any abstract container that permits the addition of new types of data? I know how to simulate the extension of Algebraic datatypes, but this does not permit the addition of data with new types in the same container and recover them in a type-safe way. Did I reinvent the Weel? I found something, that permits this for any Typeable datatype. For example x=5 list= [put x, put "hello"] [t1,t2 ]= list x' = get t1 :: Int s = get t2 :: String print (x' +1) -- 2 print s -- "hello" x''= get t2 :: Int --"get: casting from String to type Int" The code: data Abstract= forall a. Typeable a => T !String a class FromToAbstract x where put :: x -> Abstract get :: Abstract -> x unsafeGet :: Abstract -> x -- get(put x)== x instance Typeable x => FromToAbstract x where put x= T (typeString x) x get (Abstract type1 a)= if type2 == type1 then v else error ("get: casting "++ "from type "++type1++" to type "++type2) where v= unsafeCoerce a :: x type2= typeString v unsafeGet (Abstract type1 a)= unsafeCoerce a typeString !x= tyConString $ typeRepTyCon $ typeOf x instance Typeable T where typeOf _= mkTyConApp (mkTyCon "Abstract") []

3 2

Hackage web interface
by Hugo Pacheco 12 Nov '08

12 Nov '08

Hi, When previewing some package via the Hackage web interface, I get the following warning: Exposed modules use unallocated top-level names: AI assume that if you define some module A.B, it expects the top-level name A to be a module. Does this make sense? Thanks, hugo -- www.di.uminho.pt/~hpacheco

2 1

problem with boolean splicing in templates in ghc 6.10.1
by Leonidas Fegaras 12 Nov '08

12 Nov '08

Seems that boolean splicing in haskell templates in ghc 6.10.1 does not work correctly. If you do: $((\b -> [| b |]) True) you get the error: Can't find interface-file declaration for data constructor GHC.Base.True Probable cause: bug in .hi-boot file, or inconsistent .hi file Use -ddump-if-trace to get an idea of which file caused the error It seems that GHC.Base.True is now GHC.Bool.True, but the splicing code still generates the former. Leonidas Fegaras

2 1

reliable (bi)directional pipe to a process
by oleg＠okmij.org 12 Nov '08

12 Nov '08

I'd like to point out a reliable, proven and simple way of interacting with another process, via unidirectional or bidirectional pipes. The method supports Unix sockets, pipes, and TCP sockets. I too have noticed insidious bugs in GHC run-time when communicating with another process via a pipe. I tried to use runInteractiveProcess; it worked -- up to file sizes of about 300Kb. Then GHC run-time seems to `loses synchronization' -- and corrupts IO buffers, receiving stuff that cannot have been possibly sent. This is because handle operations are asynchronous and the GHC scheduler seems to have race conditions. That behavior was totally unacceptable. I was writing a production code, and can't afford such errors. Therefore, I wrote a simple foreign function interface to the code sys_open that I have been using for about fifteen years. This code does work, in production, for very large file sizes and long-running processes, on many Unix and Unix-like systems. I was told once about a Cygwin port. http://okmij.org/ftp/syscall-interpose.html#Application http://okmij.org/ftp/packages/sys_open.c http://okmij.org/ftp/Haskell/MySysOpen.hs Please see the test at the end of the file MySysOpen.hs. The test interacts with another process over a bi-directional pipe, repeatedly sending and receiving data. The amount of received data is large (about 510K).

5 6

ByteString/parsec
by Pieter Laeremans 12 Nov '08

12 Nov '08

HI, I want to efficiently parse a large collections of files. The files are in the format : example title TITLE author name AUTHOR some lines with summary here SUMMARY the real text TEXT a list of links LINKS I want to use "ByteString" here, but which library should I use to parse ? "attoparsec" or "bytestringparser", both export the same interface. When I use one of these I thaught it would be nice to write something like this : fileParser :: Parser Content fileParser = do title <- manyTill getInput (string . pack "\nTITLE\n") author <- manyTill getInput (string. pack "\nTITLE\n") .... return Content title author ... But this doesn't work. Even on a small example : parseTest (manyTill getInput (string $ pack "SPLIT") (pack "split the text at SPLIT part two") I get a stack overflow. Obviously I'm not understanding something here. Are there any good examples of open source projects which parse ByteString data ? thanks in advance, Pieter -- Pieter Laeremans <pieter(a)laeremans.org> "The future is here. It's just not evenly distributed yet." W. Gibson

3 2

compiling haskell-src-exts in 6.10
by Hugo Pacheco 12 Nov '08

12 Nov '08

When installing package haskell-src-exts via cabal install, I get the error Language/Haskell/Exts/Syntax.hs:102:7: Could not find module `Data.Data': it is a member of package base, which is hidden However, when manually installing runhaskell Setup.hs configure/build/install It works fine Somehow package base is not available in cabal.. don't know why. hugo -- www.di.uminho.pt/~hpacheco

3 2

Re[2]: [Haskell-cafe] mapping unfreeze over an IntMap of IOUArrays
by Bulat Ziganshin 11 Nov '08

11 Nov '08

Hello Max, Tuesday, November 11, 2008, 11:50:28 PM, you wrote: btw, i made here some time ago proposal about pure hashtables implented over a pure arrays (via accumArray operaion). may be it is somewhat helpful for you >>> > using unsafeFreeze. I'm getting stuck here, since the >>> IntMap library >>> > is not so monad-friendly. >>> >>> Data.Hashtable is >> >> Well, I need mutable update for a while... after that, I prefer a pure >> interface, which is why I'm trying to freeze all the values. > Is there any reason why an immutable Hashtable shouldn't be > implemented, with a pure interface, such that hashtables can be > (unsafe)Freezed into immutable hashtables? Would this be useful? > --Max -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

2 4

mapping unfreeze over an IntMap of IOUArrays
by Chad Scherrer 11 Nov '08

11 Nov '08

Hello cafe, I've hit a bit of a monadic snag here... I'm scanning a big file, building a table of statistics. I end up with something like IO (IntMap (IOUArray Int Double)) Once I've read in the whole file and built my statistics, I don't need any more updates, so I'd like to do something like IntMap (IOUArray Int Double) -> IO (IntMap (UArray Int Double)), using unsafeFreeze. I'm getting stuck here, since the IntMap library is not so monad-friendly. I could rebuild the whole thing using sequence and lists, but the table is pretty big (4.5 million keys on a first run), so I'd prefer to avoid that. Any ideas? BTW, I probably "should" be using ST for this, but I hit the usual "type s escapes" irritation and gave up. If that would work more easily, that would be fine with me too. Thanks! Chad Scherrer

5 6