August 2006 - Haskell-Cafe

Strictness, order of IO operations: NewCGI & HDBC
by Tim Smith 10 Aug '06

10 Aug '06

Hello! I'm having a problem with using NewCGI & HDBC together. NewCGI: http://www.cs.chalmers.se/~bringert/darcs/haskell-cgi/doc/ HDBC: http://quux.org/devel/hdbc http://darcs.complete.org/hdbc/ I've distilled it down to a small test case: \begin{code} module Main where import Data.List (intersperse) import qualified Database.HDBC as DB import Database.HDBC.ODBC (connectODBC) import Network.NewCGI main :: IO () main = runCGI (handleErrors cgiMain) cgiMain :: CGI CGIResult cgiMain = do dbh <- liftIO $ connectODBC "DSN=test" res <- liftIO (DB.getTables dbh) -- Remove the disconnect call, and all works liftIO (DB.disconnect dbh) output ((concat . intersperse ", ") res) \end{code} 2:46 ~/m/tmp/hs$ ./cgidb Content-type: text/html; charset=ISO-8859-1 cgidb: unknown exception If I comment out the DB.disconnect call, then it works fine. I tried to work around it with the DeepSeq module, but couldn't find an application of $!! or deepSeq which would make it function. I made a similar test which uses file IO instead of HDBC, and it behaves the way I want: \begin{code} module Main where import Network.NewCGI import System.IO main :: IO () main = runCGI (handleErrors cgiMain) cgiMain :: CGI CGIResult cgiMain = do h <- liftIO $ openFile "t.hs" ReadMode res <- liftIO (hGetLine h) liftIO (hClose h) output res \end{code} 2:55 ~/m/tmp/hs$ ./t Content-type: text/html; charset=ISO-8859-1 module Main My guess is that I'm doing something wrong, but I'm not sure what it is. Or, is this a bug in HDBC, somehow? Thanks for any ideas you might have, Tim -- If you're not part of the solution, you're part of the precipitate.

1 0

IA64 porting: spill code in the mangler
by C Rodrigues 10 Aug '06

10 Aug '06

Hi folks, I've been trying to compile a new ia64 port. I've cross-compiled an unregisterised compiler that generated working binaries the first time it was built, which was a pleasant experience. But I ran into issues with the registerised build. The mangler is choking on floating-point spill code: Prologue junk?: .global __divdf3# .global GHCziFloat_zdwlogBase_entry# .proc GHCziFloat_zdwlogBase_entry# GHCziFloat_zdwlogBase_entry: mov r16 = r12 .save.f 0x1 stf.spill [r16] = f2 .body This saves floating-point register 2 on the stack; it's loaded again later. I could change the mangler to remove this code, if that's the right approach. What should the mangler do about it?

1 0

beginner's haskell question
by Jens Theisen 09 Aug '06

09 Aug '06

Hello, as a haskell newbie I'm wondering about the following question. Are there options to popular haskell implementations or other means (haskell lint?) to check for incomplete patterns at compile time for some? I can't see a reason why this shouldn't be possible or even a relatively simple thing to implement. Cheers, Jens

3 3

Re: Haskell-Cafe Digest, Vol 36, Issue 29
by Mark Haniford 09 Aug '06

09 Aug '06

It seems irony gets lost so easily in these conversations. You have no > way of knowing what the state of haskell would have been had certain > key contributors to GHC and Haskell not taken jobs at Microsoft. > Therefore you statement is meaningless and only good for producing > approving nods among people who already agree with what you say. What's meaningless are GPL jihadists who are bitter that not everybody is joining their religion. Yeah...that's personal.

1 0

Why Not Haskell?
by Hans van Thiel 09 Aug '06

09 Aug '06

Hello All, I'm wondering why I can't find any commercial Haskell applications on the Internet. Is there any reason for this? I can think of the following possibilities only: 1) Haskell is too slow for practical use, but the benchmarks I found appear to contradict this. 2) Input and output are not good enough, in particular for graphical user interfacing and/or data base interaction. But it seems there are several user interfaces and SQL and other data base interfaces for Haskell, even though the tutorials don't seem to cover this. 3) Haskell is not scaleable for commercial use. This looks unlikely to me, but could this be a factor? 4) Haskell is open source and licensing restrictions forbid commercial applications. I haven't seen any such restrictions, but is this a problem for the standard modules? Are there other reasons why there seem to be just a few thousand (hundred?) Haskell programmers in the world, compared to the 3 million Java programmers and x million C/C++ programmers? The reason I ask is that in the early nineties I wrote a data mining /machine learning program. This was not a commercial success and since then I've been working as a free lance IT journalist. As such I covered the O'Reilly European Open Source Convention in Amsterdam last year and there I heard of Haskell. I became interested and started learning from the available free tutorials and Hugs. As an aside, I'd like to thank all those people. I really appreciate their work and I'd like to single out Hal Daume III's 'Yet Another Haskell Tutorial', while not diminishing the other authors in any way. Meanwhile I kept working on a data mining algorithm in my spare time and finally I came up with something that appears to work. While I was learning Haskell, I thought, why not try this part of the algorithm as a Haskell exercise. I could easily test it in Hugs. To my own surprise, after only a few weeks of this, I had a working prototype of my algorithm, printed out on less than 3 pages, using only standard functions in the Prelude. So, I'm really impressed by the power of Haskell, and I'm just a beginner. Now I'm trying to come up with a business model for my algorithm and to avoid the mistakes I made 10 years ago. There is a lot of difference between a prototype and a working tool, and then there is a lot of difference between a working tool and a successful commercial application. Probably it doesn't make much sense to try and develop a tool in C++ or even Java, but if I have to go on my own on this, maybe Haskell could be feasible, both for fun and profit. So, that's the background of my question (while introducing myself to the Cafe at the same time). Why hasn't Haskell made it into the business world (yet), after being available for 15 years, or is this the wrong question? Many thanks for your comments, Hans van Thiel

24 45

HWS - With Plugins
by Johan Tibell 09 Aug '06

09 Aug '06

The HWS - With Plugins tarball is unavailable at the author's website (http://www.mdstud.chalmers.se/~md9ms/hws-wp/) and his email address doesn't work so this is desperate attempt to reach him. So, Martin Sjögren, are you here somewhere? P.S. If someone else knows where I could get hold of the source or at least the source of the original HWS I would be greatful. Cheers, Johan

2 1

Abstract Data Types
by Johan Grönqvist 09 Aug '06

09 Aug '06

Hi, I have a question: Short version: If I want to hide the implementation of a data-type "Stack a" from the rest of the program, do I need to put its definition in a separate file? Long version: I want to use a stack, and I might implement it as a list, but I want to hide the implementation from the rest of the program. This is how I understand abstract data type. In "The Craft of Functional Programming", this seems to be implemented by putting each data type into a separate module and only exporting parts of the definitions. In "The Haskell School of Expression", this seems not to be used at all. In the lecture notes at (http://www.dcs.shef.ac.uk/~mps/courses/com2020/adts.pdf) type classes are used for abstract data types. It seems to me that this approach does not hide any parts of any definition, but only requires that all instances of class stack have functions pop and push of the correct types. I am interested in hiding parts of definitions. In the report, I did not find any mention of a requirement to have different modules in separate files, but I have not managed to put several modules in the same file using ghci. I would like to keep my small program in one literate-haskell tex-file and still be able to hide some definitions from others. Is this possible? One option would of course be to write a script that separates the code into different and then compiles the entire program. Thanks in advance! / johan grönqvist

2 1

AJAX applications in Haskell
by tittoassini＠gmail.com 09 Aug '06

09 Aug '06

Hi, Those among you who have an interest in AJAX-style web development - that is to say the development of web applications that run entirely into the browser environment, calling back to the server back-end only to get raw data -- will probably have noticed the recent appearance of the Google Web Toolkit (http://code.google.com/webtoolkit) The main idea behind it is very simple: you write your Ajax application in Java and your code is compiled down to JavaScript/HTML (rather then to Java Virtual Machine bytecode) so that it can be executed in any browser. The main advantages of this solution are: - Development using a typed language with good compile-time checks and debugger support - No plugins (e.g. a Java VM) required on the user machine The disadvantages are: - An additional compilation step - Probably, a significant loss of performance with respect to hand-written code Assuming that the balance is positive, this naturally raises the question: why not doing the same with our favourite language? This would require to: - Retarget one of the existing Haskell compilers to generate JavaScript (other possible targets would be Flash or higher level UI languages such as OpenLaszlo that in turn compiles down to either Flash or JavaScript/HTML) - Write a suitable runtime library (including support for native JavaScript or Flash UI components, remote method invocations, multi-threading, etc.) - Port a usable subset of the Haskell Libraries - Select an approach to write user interfaces in Haskell. There have been many attempts in this direction but I am unsure if any of them has proven particularly successful. All considered, quite a significant amount of work -- that would require a concerted group effort -- with a rather uncertain pay-off. Is anyone working on anything similar or that might be interested in such a project? And has anyone any thoughts to offer on what compilers/libraries/UI frameworks would be more suitable for the task? Regards, titto assini

1 0

Re: [Haskell-cafe] creating tree with level subnodes and incrementing number?
by Marc Weber 08 Aug '06

08 Aug '06

Thanks to you all! I think I've learned a lot. Marc

1 0

IO Monad/ haskelldb strange error?
by Marc Weber 08 Aug '06

08 Aug '06

line << 39 works fine line 56 doesn't. Why? Isn't both a IO monad (because of the print statements) After commenting out 56 it compiles fine Any suggestion appreciated.. I'm struggling for some hours now.. module Modules.ObjectTree where import Debug.Trace import Data.FunctorM import DBUtils import qualified DB.VT.Ezcontentobject_tree as EOT import qualified DB.VT.Ezcontentobject as CO import Database.HaskellDB.HDBRec import Database.HaskellDB import Database.HaskellDB.Query as Q import Data.Tree import Monad import Control.Monad.Trans import Maybe import qualified List instance FunctorM Tree where fmapM f (Node a forest) = do a' <- f a forest' <- mapM (fmapM f) forest return $ Node a' forest' type ObjectTree a = Tree (Record a) truncTree 1 (Node a _) = Node a [] truncTree x (Node a forest) = Node a $ map (truncTree (x-1)) forest oT con = do print "blah" -- because of this we should have a simple IO Monad ? lookupField con (CO.name) (CO.ezcontentobject) (CO.xid) (constant (1 :: Int)) >>= print :: IO () -- <<<<<<<<<<<<<<<<<<<<<<<<< 39 return "blah" -- printObjectsAsTree :: MonadIO m => ((Database -> m a) -> m a) -> Int -> IO () printObjectsAsTree con startid= do print "test" root <- liftM head $ lRS (EOT.parent_node_id) (constant (startid :: Int)) print root --showRS root >>= putStrLn node <- po root node_show <- fmapM showRS node return $ drawTree node_show -- return "end" where lRS = lookupFieldRS con (EOT.ezcontentobject_tree) po root = let root_id = (root!(EOT.node_id) :: Int) in do print "dumm" -- IO Monad too ? print (root!(EOT.node_id)) lookupField con (CO.name) (CO.ezcontentobject) (CO.xid) (constant (1 :: Int)) >>= print -- <<<<<<<<<<<<<< 56 return $ Node root [] --childs <- lRS (EOT.parent_node_id) (constant root_id) --mapM_ (\r -> r!(EOT.node_id)) childs >>= print showRS r = do -- name <- lookupField con (CO.name) (CO.ezcontentobject) (CO.xid) (constant 1) >>= print return "ab" :: IO String --return $ (show $ r!node_id) ++ " (" ++ (fromJust name) ++ " )" ----------------------------------------------- || Preprocessing executables for dbez-0.0... || Building dbez-0.0... || Chasing modules from: db_ez.hs || [1 of 6] Skipping DBUtils ( DBUtils.hs, dist/build/db_ez/db_ez-tmp/DBUtils.o ) || [2 of 6] Skipping DB.VT.Ezcontentobject_tree ( DB/VT/Ezcontentobject_tree.hs, dist/build/db_ez/db_ez-tmp/DB/VT/Ezcontentobject_tree.o ) || [3 of 6] Skipping DB.VT.Ezcontentobject ( DB/VT/Ezcontentobject.hs, dist/build/db_ez/db_ez-tmp/DB/VT/Ezcontentobject.o ) || [4 of 6] Compiling Modules.ObjectTree ( Modules/ObjectTree.hs, dist/build/db_ez/db_ez-tmp/Modules/ObjectTree.o ) || Modules/ObjectTree.hs|43| 0: || Couldn't match `DB.VT.Ezcontentobject_tree.Contentobject_id' || against `DB.VT.Ezcontentobject.Contentclass_id' || Expected type: RecCons DB.VT.Ezcontentobject_tree.Contentobject_id || (Maybe Int) || vr || Inferred type: RecCons DB.VT.Ezcontentobject.Contentclass_id || Int || vr1 || When using functional dependencies to combine || Database.HaskellDB.Database.GetRec (RecCons f (Expr a) er) || (RecCons f a vr), || arising from the instance declaration at Imported from Database.HaskellDB.Database || Database.HaskellDB.Database.GetRec (RecCons DB.VT.Ezcontentobject_tree.Contentobject_id || (Expr (Maybe Int)) [...] || (RecCons DB.VT.Ezcontentobject_tree.Sort_order || (Expr (Maybe Int)) || RecNil)))))))))))))))) || (RecCons DB.VT.Ezcontentobject.Contentclass_id Int vr), arising from use of `lookupFieldRS' at Modules/ObjectTree.hs|52| 14-26 || When generalising the type(s) for `printObjectsAsTree'

4 4