- Haskell - Haskell.org

ANNOUNCE: nhc98 1.04 and Hat 1.04
by Malcolm Wallace 21 May '01

21 May '01

We are pleased to announce a new release of the nhc98 compiler, and in particular its facilities (both brand new and improved old stuff) for tracing and debugging Haskell programs - the Hat system. The basic nhc98 compiler version 1.04 is mostly a bugfix release (details listed at the bottom of this announcement). http://www.cs.york.ac.uk/fp/nhc98/ However, the stuff we are really excited about is Hat - our Haskell tracer. A full user guide to these new tools is available on the website. http://www.cs.york.ac.uk/fp/hat/ Hat 1.04 ======== * Lots more tools for tracing! hat-stack - Displays a "virtual" stack-trace for any program that failed with a runtime error. hat-detect - Algorithmic bug detection, based on a textual question-and-answer session. hat-trail - The original graphical trail browser, now with several presentation enhancements. hat-observe - Examine the inputs and outputs of any named function, in the style of HOOD. * Tracing with Hat is now based on a completely new architecture (although still within the nhc98 compilation system). A program compiled for tracing now builds its "redex trails" in file, not in the heap. This has some performance consequences: (1) you no longer need to allocate large amounts of heap memory for tracing; (2) but you will probably need large amounts of disk space instead - remember disk space is cheap; (3) and traced programs still run somewhat slowly. * However, there are a couple of major benefits in return: (1) A trace is persistent, so after running your program once, you can examine many different aspects of the trace without re-computation. (2) Traces are now first-class objects, so you can examine and manipulate them in far more detail - hence the variety of new tracing tools. * Greater Haskell'98 compatibility. Almost all language features are now supported: named fields and better handling of pattern bindings are the main additions. All standard libraries except Time and Locale are now supported. nhc98 1.04 ========== * New: Improved (more accurate) time profiling now provided. * New: Support for extended module namespaces of the form Long.Hierarchical.Module.Name is now provided in both nhc98 and hmake. * Bugfix: An identifier hidden on import and redefined in the current module, then exported, but also imported qualified and used qualified in the current module, led to an incorrect interface file being generated. * Bugfix: hmake issued an unnecessary -cpp flag on some literate files. * Bugfix: Type of IO.hSetPosn :: Handle -> HandlePosn -> IO () was incorrect. * Bugfix: Compile-time error in <tt>src/tracer/runtime/ident.c</tt> on RedHat 7 and other systems using the new ISO C standard for <tt>fpos_t</tt>. * Bugfix: A file opened in ReadMode or WriteMode was actually opened in ReadWriteMode, so if the file had strict permissions the correct opening command would fail. Conversely, opening in ReadWriteMode actually gave ReadMode instead, and file updates silently failed. * Bugfix: Operator sections suffered from priority inversion, for example (^2*3) was incorrectly parsed as (^(2*3)), even though ^ binds more tightly than *. * Bugfix: The library function Directory.createDirectory gave strange permissions to the new directory. (Mode was in hex, but should have been octal!) * New: Improved printing of I/O error messages. Happy tracing! The Hat team at York (Colin Runciman, Olaf Chitil, Malcolm Wallace, Thorsten Brehm, Phil Hassall)

1 0

Re: PP with HaXml 1.02
by Ketil Malde 21 May '01

21 May '01

"Taesch, Luc" <luc.taesch(a)csfb.com> writes: > the pretty print return the closing < on the next line, like > <Person-XML > ><Name > ><string .... > where i expect > <Person-XML> > <Name> > <string> My guess is that HaXML doesn't validate (using the DTD) or simply ignores it for pretty printing purposes. If the "Person-XML" element were defined to hold #PCDATA as well as (a) "Name" element(s), the two examples above are not equivalent, since you've added white space characters to the contents. Adding whitespace within the tag is okay, though, since it doesn't change the semantics of the document. > This message is for the named person's use only. It may contain [...] (Jeez! Has CSFB hear of mailing lists?) -kzm -- If I haven't seen further, it is by standing in the footprints of giants

1 0

PP with HaXml 1.02
by Taesch, Luc 21 May '01

21 May '01

Im proceeding with basic exploratory tests witrh HaXml (1.02, winhugs feb 2001) the pretty print return the closing < on the next line, like <Person-XML ><Name ><string .... where i expect <Person-XML> <Name> <string> what am i doing wrong ? ( the call is like import qualified XmlPP as PP ... main = do content <- readFile "subjdb.xml" (putStrLn . render . PP.document . parse) content (btw, thanks malcolm for packaging a specific hugs distribution) This message is for the named person's use only. It may contain confidential, proprietary or legally privileged information. No confidentiality or privilege is waived or lost by any mistransmission. If you receive this message in error, please immediately delete it and all copies of it from your system, destroy any hard copies of it and notify the sender. You must not, directly or indirectly, use, disclose, distribute, print, or copy any part of this message if you are not the intended recipient. CREDIT SUISSE GROUP and each of its subsidiaries each reserve the right to monitor all e-mail communications through its networks. Any views expressed in this message are those of the individual sender, except where the message states otherwise and the sender is authorised to state them to be the views of any such entity. Unless otherwise stated, any pricing information given in this message is indicative only, is subject to change and does not constitute an offer to deal at any price quoted. Any reference to the terms of executed transactions should be treated as preliminary only and subject to our formal written confirmation.

1 0

Proposal for generalized function partition in List-library
by Bernd Holzmüller 19 May '01

19 May '01

Hi all, knowing that the Haskell library report is currently being rewritten, I would like to propose a new function for module List that generalizes the current function partition :: (a -> Bool) -> [a] -> [[a]] in the following way: partition :: Eq b => (a -> b) -> [a] -> [[a]] partition _ [] = [] partition f (a:as) = let (as',as'') = foldr (select (f a)) ([],[]) as in (a:as'):partition f as'' where select b x (ts,fs) | f x == b = (x:ts,fs) | otherwise = (ts,x:fs) This partitioning function builds equivalence classes from the list argument, where each element list within the result list consists of elements that all map to the same value when applying f to it. Thus: partition (`div` 2) [1..5] yields [[1],[2,3],[4,5]] This is much more general than the existing partitioning function and applicable in many practical cases. Cheers, Bernd Holzmüller

11 14

Re: Templates in FPL?
by George Russell 18 May '01

18 May '01

The MLj compiler from ML to the Java Virtual Machine (which is being actively worked on by the geniuses at Microsoft as we speak) expands out all polymorphism. I helped develop this compiler and I believe this approach to be a good one. There are particular reasons why it's good for this case (1) as has been said before, ML does not have polymorphic recursion; (2) the cost of object creation in Java used at least to be horrendous, so it was well worth working hard to get rid of it. However I think long term it's a good idea for normal Haskell compilers because my guess is you will Win Big on quite a lot of things, because you don't have to carry around and use dictionaries (for method lookup in class instances) and can do a lot more specialisation and inlining. Strict values can also be unboxed. Of course you can control specialisation used unboxed values already in Glasgow Haskell, but it requires special annotations or special types. The minus points are thuswise, so far as I know: (1) the compiled code is bigger. This means the code might indeed run more slowly. But with MLj I think we found that in normal cases it isn't THAT much bigger (maybe about 50%), because the functions which get specialised a lot (like "length" and "map") tend to be pretty small. Of course you can construct artificial cases where the code size will explode. (2) It takes much longer to compile. I'm waiting for Moore's law here, but at present I would be surprised if (say) Glasgow Haskell could compile itself inlining out all polymorphism, without it taking a very long time or a very fast computer or lots of nasty imperative optimisations. (3) You can't inline everything, because of polymorphic recursion, and also because it would be nice to allow Haskell code to be dynamically linked in (see GHCi). Or you can inline everything, but then you need to implement a Haskell Virtual Machine which does the inlining (and the whole compiler backend) on demand at runtime. I don't think you have to inline everything; it should surely be possible to leave polymorphic versions of the functions around, to use for hard cases.

2 1

Monads
by Mads Skagen 18 May '01

18 May '01

Thank you for your reply. So what you are saying is that I actually don't need Monads to perform the tasks Monads supports ? Thank you very much. Regards Skagen ______________________________________________________ Går mail for langsomt for dig? Så prøv Yahoo! Messenger - her kan du i løbet af få sekunder udveksle beskeder med de venner, der er online. Messenger finder du på adressen: http://dk.messenger.yahoo.com

5 4

MVar Problem (Concurrent Hugs)
by Andreas Gruenbacher 18 May '01

18 May '01

Hello, I was trying to write an abstraction for bidirectional communication between two threads. For some reason, MVars seem to break: ----------------------------------------------- class Cords c t u where newCord :: IO (c t u) listen :: c t u -> IO t speak :: c t u -> u -> IO () data Cord t u = Cord (IO (MVar t)) (IO (MVar u)) instance Cords Cord t u where newCord = return (Cord newEmptyMVar newEmptyMVar) speak (Cord _ s) t = do s' <- s ; putMVar s' t listen (Cord h s) = do h' <- h ; takeMVar h' otherEnd (Cord t u) = Cord u t showT :: Cord Int String -> IO () showT cord = do putStrLn "Runnning..." x <- listen cord putStrLn ("Heard " ++ show x) speak cord (show x) putStr ("Said " ++ (show x)) showT cord main :: IO () main = do cord <- newCord forkIO (showT (otherEnd cord)) speak cord 1 str <- listen cord putStrLn str ----------------------------------------------- Hugs writes: Main> main Runnning... Program error: no more threads (deadlock?) (131 reductions, 307 cells) Any ideas? Thanks, Andreas. ------------------------------------------------------------------------ Andreas Gruenbacher gruenbacher(a)geoinfo.tuwien.ac.at Research Assistant Phone +43(1)58801-12723 Institute for Geoinformation Fax +43(1)58801-12799 Technical University of Vienna Cell phone +43(664)4064789

2 2

re: Questions about Trie
by Colin.Runciman＠cs.york.ac.uk 18 May '01

18 May '01

> Greeting, I'm a last year student in a computer science field. I'm > currently trying to code an implementation for a compression using basic > Lempel-zif technique. I use a Trie (retrieval) as a data structure for > the dynamic dictionary aplication. The problem is Trie uses not just an > ordinary binary tree, but it uses a multiple-weighted tree, where one > node has at most 256 children. Most literatures about Haskell show only > binary tree for the example in tree or Abstract Data Structures subject. > So, does anyone know how to code the implementation for trie. Thanks for > paying attention and I'm really hoping for the answer immediately. There is some relevant material in Chapter 6 (Text compression) of `Applications of Functional Programming', UCL Press, 1995. ISBN: 1-85728-377-5. (With apologies for self-reference, as co-editor of the book.) Colin R

1 0

RE: Questions about Trie
by Simon Peyton-Jones 18 May '01

18 May '01

Perhaps this monads discussion might move to haskell-cafe@haskell.org? It's a good discussion, but it's just what haskell-cafe is for. http://www.haskell.org/mailinglist.html Simon | -----Original Message----- | From: Arthur H. Gold [mailto:agold@bga.com] | Sent: 17 May 2001 22:21 | To: haskell(a)haskell.org | Cc: M. Faisal Fariduddin Attar Nasution | Subject: Re: Questions about Trie | | | > "M. Faisal Fariduddin Attar Nasution" wrote: | > | > Greeting, I'm a last year student in a computer science | field. I'm | > currently trying to code an implementation for a | compression using | > basic Lempel-zif technique. I use a Trie (retrieval) | as a data | > structure for the dynamic dictionary aplication. The | problem is Trie | > uses not just an ordinary binary tree, but it uses a | multiple-weighted | > tree, where one node has at most 256 children. Most | literatures about | > Haskell show only binary tree for the example in tree or | Abstract Data | > Structures subject. So, does anyone know how to | code the | > implementation for trie. Thanks for paying attention and | I'm really | > hoping for the answer immediately. | > | | IIRC, there's a trie implementation in Chris Okasaki's | "Purely Functional Data Structures" -- that might be a start | (there's also his "Edison" library, which contains many of | the components you'd need). | | HTH (and sorry if it doesn't) | --ag | | -- | Artie Gold, Austin, TX (finger the cs.utexas.edu account for | more info) mailto:agold@bga.com or mailto:agold@cs.utexas.edu | -- | I am looking for work. Contact me. | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org http://www.haskell.org/mailman/listinfo/haskell |

1 0

Questions about Trie
by M. Faisal Fariduddin Attar Nasution 17 May '01

17 May '01

Greeting, I'm a last year student in a computer science field. I'm currently trying to code an implementation for a compression using basic Lempel-zif technique. I use a Trie (retrieval) as a data structure for the dynamic dictionary aplication. The problem is Trie uses not just an ordinary binary tree, but it uses a multiple-weighted tree, where one node has at most 256 children. Most literatures about Haskell show only binary tree for the example in tree or Abstract Data Structures subject. So, does anyone know how to code the implementation for trie. Thanks for paying attention and I'm really hoping for the answer immediately. Aerith Gainsborough (Final Fantasy VII) "All I know is..." "The Cetra were born from the Planet, speak with the Planet, and unlock the Planet." "And....... then......" "The Cetra will return to the Promised Land. A land that promises supreme happiness." =================================== M. Faisal Fariduddin Attar Nasution ----------------------------------- farkinas(a)cbn.net.id

2 1