October 2003 - Glasgow-haskell-users

RE: Transmitting Haskell values
by Simon Peyton-Jones 29 Oct '03

29 Oct '03

| > The GHC versions are simply more efficient versions. Show and Read are | > completely portable. | | AIIEE! Now they tell me -- I've optimized one of my programs, and it | now spends most of it's time 'read'ing. So the thing is just to | import GHC(read) instead of using the Prelude version? I guess it's | worth a try (although I expect I really need to go to binary IO) No, you get the efficient ones when you import the Prelude. GHC's Read class has two extra methods, readPrec and readListPrec, which read using a parser monad rather than the 'list of successes' that is (sadly) built into the definition of the standard Read class. When defining a new type, and making a Read instance for it, you'll get more efficient reading if you define GHC's Read methods (the normal readsPrec and readList will then be defined in terms of them). There are many examples in GHC.Base.Read The underlying parser technology is Koen Claessens's parallel parser http://www.cs.chalmers.se/~koen/publications.html This new parsing stuff is hardly documented at all (e.g. http://www.haskell.org/ghc/docs/latest/html/base/GHC.Read.html#readPrec) . If anyone would like to remedy this by writing Haddock documentation, it'd be great. (I'd be glad to hand-hold.) Simon

1 0

Templates...
by MR K P SCHUPKE 29 Oct '03

29 Oct '03

I would like to create a template to generate type declarations, for example: newtype "mytype" = "myconstructor" String I am struggling without an example... how do I do it? Regards, Keean Schupke.

1 0

unknown symbol `__stginit_List_'
by Josef Svenningsson 29 Oct '03

29 Oct '03

Hi! Here's a letter from someone who has next to no clue about what he is doing so have patience and don't assume any knowledge when answering... I'm trying to create a package for ghc. I've struggled quite a lot and gotten this far: (This is when invoking ghci -package yahu) Loading package base ... linking ... done. Loading package yahu ... linking ... /.../chalmers.se/fs/cab/cs/work/proj/multi/pub/lib/yahu/Yahu/YahuHaskell.o: unknown symbol `__stginit_List_' ghc-6.0.1: panic! (the `impossible' happened, GHC version 6.0.1): can't load package `yahu' I'm surprised by this. Why do I get this error message? From the name of the symbol I would expect that it can be found in package base but aparently not... Please advice. /Josef

3 3

GHC changes
by Steffen Mazanek 29 Oct '03

29 Oct '03

Hello GHC freaks, I have made some changes to the GHC source code (6.0.1) and I would really welcome your feedback, whether they are ok for my task (first step: introduce record expressions of the shape 'struct {x1=e1,x2=e2,...,xn=en}'): * added ("struct", ITstruct, bit glaExtsBit), to ghcExtensionKeywordsFM and ITstruct to Token in Lex.lhs * added | Struct [(id, HsExpr id)] to data type HsExpr id in HsExpr.lhs Please tell me, what is this id for? Can it be used in this context? What is FastString? * added 'struct' { ITstruct } to the token part of Parser.y (please tell me, why in the lexer we declare ITstruct to represent the keyword "struct" and in the parser specification we turn back to 'struct'), added structassignments : structassignments ',' structassignment { $3 : $1 } | structassignments ',' { $1 } | structassignment { [$1] } | {- empty -} { [] } structassignment : qvar '=' exp { ($1,$3) } as new production rules (what type has structassignments? I can not simply write [(id,HsExpr id)], is qvar '=' exp appropriate? I do not understand structassignments completely (I have tried to make it analog to fbinds)), extended exp10 by | 'struct' '{' structassignments '}' {Struct $3} | '{' structassignments '}' {Struct $2} and finally added | Struct [(id, HsExpr id)] to HsExpr How can I speed up the recompilation process? I just want to check, whether my changes are compilable or not, is it possible to switch off optimization and other time consuming transformations? At next I will try to master the following steps: * embed the remaining new production rules to the parser, e.g., record type declaration, selectors, particular kind annotations * adaptation of the type checker * reduction of struct expressions Thank you very much, Steffen Mazanek

1 0

File permissions of /tmp files
by Peter Simons 29 Oct '03

29 Oct '03

Hi, it appears that ghc (and ghci) create temporary files in /tmp during the un-lit phase, at least. It would be nice, if ghc would ... - create those files with read/write permission for the owner only, 0600. - set-up some atexit(3) job, or whatever, to make sure those files _are_ deleted when ghc exits. It might even be possible to create the file and then remove(3) it, right away. The file handle will continue to work, but there is never an entry in the file system. Just a suggestion. :-) Peter

3 2

ghc-6.0.1 on Mac OS X 10.3 (Panther)
by Gregory Wright 28 Oct '03

28 Oct '03

Hi Wolfgang, I just updated my powerbook to Panther and tried my old ghci image (built under Jaguar). It seems to run OK, although I haven't tested it extensively. I used it with both the terminal interface and emacs. It may be possible to compile a Panther compiler using a ghc built under Jaguar; I'll test this if I don't run into other issues in the next few days. Best Wishes, Greg Wright

1 0

Special Invitation :-) HC&A Report (November 2003)
by C.Reinke 28 Oct '03

28 Oct '03

Dear GHC/GPH/GDH users and developers, in case you haven't seen the calls for contributions on the main Haskell list: the Haskell community hopes to hear from you about all the interesting stuff you've been brewing over the last six months, not to mention the even more interesting stuff in the pipeline!-) Simon PJ has sent an update for GHC itself, but there have been several interesting developments, here on the list, or on workshops, etc. during the last six months, which are not yet represented/ documented in the draft of our report. Here's a non-exclusive list of topics that are not yet included: - Simon M.' usual batch (he's away;-): - Concurrent Haskell - Alex/Happy - Haddock - hierarchical libraries (the libraries list members are taking care of this one) - Robert Ennal's new ghc debugger - Wolfgang Thaller's bound threads work - the whole parallel and distributed Haskell story.. - etc., etc., etc. .. Check the topics and contributors list to see what still needs filling in: http://www.haskell.org/communities/topics.html ----------------------------------------------------------------- Haskell Communities and Activities Report (November 2003 edition) http://www.haskell.org/communities/ Could everybody send in their individual reports ** by the end of THIS week ** ? Please?-) ----------------------------------------------------------------- Thanks, Claus (current editor)

1 0

DiffArray Performance
by Adrian Hey 28 Oct '03

28 Oct '03

Hello, I've been trying to optimise the following code.. -- Search Tree data type newtype STree = STree (Array Int (STree,[Match])) -- Initial value for Search Tree sTree0 :: STree sTree0 = STree (array (0,9) [(n,(sTree0,[]))| n <- [0..9]]) -- Make the search tree from a list of words makeSTree :: [String] -> STree makeSTree ws = foldl' putWord sTree0 pairs where pairs = [let ps = packString w in ps `seq` (word2keys w, MatchW ps) | w<-ws] word2keys cs = [getKey (toUpper c) | c <- cs, c /= '"' , c /= '-' ] putWord stree (keys,m) = put keys stree where put [] _ = error "makeSTree: empty Keys" put [k] (STree a) = let (t,ms) = a ! k in STree (a // [(k,(t,m:ms))]) put (k:ks) (STree a) = let (t,ms) = a ! k t' = put ks t in t' `seq` STree (a // [(k,(t',ms))]) This seems to be taking about 4.8 seconds (of 5.1 seconds total program execution time) for the input I'm using. I thought using DiffArrays might be faster, but no such luck. Execution time rose to 9.5 *minutes*. Is this what I should expect to see? I'm using ghc 6.0, BTW. Thanks -- Adrian Hey

2 3

command line option --show-iface
by Steffen Mazanek 27 Oct '03

27 Oct '03

Hello, the following error is slightly confusing: ghc --show-iface ghc-6.0.1: unrecognised flag: --show-iface Usage: For basic information, try the `--help' option. ghc --show-iface Unify.hi __interface "Main" Unify 1 where __export Unify Unifiable{mgu} match mguType varBind; ... A hint in the right direction would be more convenient :-) Maybe in /ghc/compiler/main/DriverUtil.hs a function unknownFlagErrUsage :: String->String -> a unknownFlagErrUsage f specusage = throwDyn (UsageError ("unrecognised flag: " ++ f ++ "\n" ++ "hint: " ++ specusage)) and changing processOneArg from DriverFlags.hs HasArg fio -> if rest /= "" then fio rest >> return args else case args of [] -> unknownFlagErrUsage dash_arg "argument for flag expected" (arg1:args1) -> fio arg1 >> return args1 Would this be ok? Regards, Steffen

1 0

Re: Question about profiling in GHC...
by Kirsten Chevalier 27 Oct '03

27 Oct '03

On Mon, Oct 27, 2003 at 12:00:04PM -0500, glasgow-haskell-users-request(a)haskell.org wrote: > When I compile my program without "-prof -auto-all" option (no > profiling support), its execution time is about 140s (compiled with > -O2). When compiled with profiling support, the time spent by the > program is about 180s (I used my own timer to measure this). Of > course, the additional 40s is caused by the profiling annotation code. > However, the profiler says (in the ".prof" file produced at the end of > execution) that the time spent is about 85s. I suppose that the time > measured by the profiler is only for evaluation of the main function > of my program (I have compiled all modules with "-prof-all"). But what > kind of computation is performed in the rest 95s (180s - 85s) ????? > Garbage collection ??? Yes, it's probably garbage collection. To be sure, you can run your program with the "-t" RTS option, which will create a file in the current working directory named "foo.stat" if the executable is named "foo". The resulting file will contain the total amount of time spent, the mutator time, and the GC time. (I recently ran into this problem myself...) -- Kirsten Chevalier * krc(a)cs.berkeley.edu * Often in error, never in doubt "But just because we're conditioned to view some things as disgusting and immoral doesn't mean that some things aren't, in actual point of fact, disgusting and immoral. Human sacrifice, for instance. Or cannibalism. Or Ann Coulter." -- Dan Savage http://www.cs.berkeley.edu/~krc/

1 0