February 2004 - Haskell

Re: [Haskell] Weaving the Web with Haskell
by Graham Klyne 04 Feb '04

04 Feb '04

If I can get the POpen module working under Windows, we have a plausible route to achieving this using existing XSLT programs. See my message: http://haskell.org/pipermail/haskell/2004-February/013522.html Indeed, this could be a powerful tool for adding functionality quickly to Haskell on the major platforms (*nix and Windpows). Any help with this would be appreciated! #g -- At 20:55 03/02/04 -0700, Seth Kurtzberg wrote: >Don't forget about XSLT translation. In the commercial projects that I >have worked recently, using XSLT to translate to printed and screen form >representations. > >If I can help with this aspect, or any other aspect of the project, please >let me know. ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact

1 0

RE: [Haskell] RE: Hugs/GHC incompatibility
by Simon Peyton-Jones 04 Feb '04

04 Feb '04

Yes, that's it. We don't want an extra conditional test on every store, plus perhaps an array of bits to record which items have been filled. Simon | -----Original Message----- | From: haskell-bounces(a)haskell.org [mailto:haskell-bounces@haskell.org] On Behalf Of Iavor S. | Diatchki | Sent: 03 February 2004 18:23 | To: Wolfgang Jeltsch | Cc: The Haskell Mailing List | Subject: Re: [Haskell] RE: Hugs/GHC incompatibility | | hi, | i am speculating here, but i guess that if double initializations were to | result in _|_ one would have to check if a cell has already been | initialized, | and if so replace it with _|_. | currently ghc probably simply overwrites the array elements. | i personally prefer ghc's behavior. | -iavor | | Wolfgang Jeltsch wrote: | | >Am Dienstag, 3. Februar 2004 18:17 schrieb Simon Peyton-Jones: | > | > | >>>I think, I've found the problem. GHC seems to not implement | >>>Data.Array.array correctly. The implementation doesn't adhere to the | >>>following sentence from the Data.Array.IArray.array documentation: If any | >>>two associations in the list have the same index, the value at that index | >>>is undefined. | >>> | >>>Could this be fixed? | >>> | >>> | > | > | > | >>[...] | >> | >> | > | > | > | >>We don't intend to fix this because fixing it will have a significant | >>overhead on every array construction. | >> | >> | > | >Sorry, but I don't see where the overhead should come from. | > | > | > | >>[...] | >> | >> | > | > | > | >>Simon | >> | >> | > | >Wolfgang | > | >_______________________________________________ | >Haskell mailing list | >Haskell(a)haskell.org | >http://www.haskell.org/mailman/listinfo/haskell | > | > | | | -- | ================================================== | | Iavor S. Diatchki, Ph.D. student | | | Department of Computer Science and Engineering | | | School of OGI at OHSU | | | http://www.cse.ogi.edu/~diatchki | | ================================================== | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/haskell

1 0

ANNOUNCE: HaskellDB 0.5
by Anders Höckersten 03 Feb '04

03 Feb '04

We are pleased to announce the first alpha release of our "remake" of HaskellDB. HaskellDB is a Haskell library for expressing database queries and operations in a type safe and declarative way. HaskellDB compiles a relational algebra-like syntax into SQL, submits the operations to the database for processing, and returns the results as ordinary Haskell values. HaskellDB uses HSQL to interface with standard database drivers, and is thus compatible with several standard databases (at the moment of writing, MySQL (only versions >=4.1), PostgreSQL and most ODBC databases). The original version of HaskellDB was written by Daan Leijen and is available at http://www.haskell.org/haskellDB/ This alpha release works pretty well so far, but has not been tested much, so we expect it to contain a lot of bugs. If you try it out and find something you suspect to be a bug, please contact us and we will try to help you as much as we are able. We also welcome all feature requests and suggestions for improvements. For more information and download see: http://haskelldb.sourceforge.net/ The developers can be reached at haskelldb-users(a)lists.sourceforge.net or on #haskelldb at irc.freenode.net. Thanks to Daan Leijen for his original version of HaskellDB. We would also like to thank Krasimir Angelov for all his help with HSQL, and for being nice enough to release a new version of it just in time for our release. Also, a special thank you to John Hughes for helping us out with everything regarding this project. Happy hacking, and don't try this on any production databases! :) / The HaskellDB development team

1 0

RE: Hugs/GHC incompatibility
by Simon Peyton-Jones 03 Feb '04

03 Feb '04

| I think, I've found the problem. GHC seems to not implement Data.Array.array | correctly. The implementation doesn't adhere to the following sentence from | the Data.Array.IArray.array documentation: | If any two associations in the list have the same index, the value at that | index is undefined. | Could this be fixed? It's a documented incompatibility with H98 http://www.haskell.org/ghc/docs/latest/html/users_guide/bugs-and-infelic ities.html#HASKELL98-DIVERGENCE We don't intend to fix this because fixing it will have a significant overhead on every array construction. Actually I rather think H98 should have left the behaviour of double indices undefined, rather than defining it to be bottom. But it's too late now. Simon

3 2

RE: [Haskell] Weaving the Web with Haskell
by Graham Klyne 03 Feb '04

03 Feb '04

At 16:23 03/02/04 +0000, Simon Peyton-Jones wrote: >Having solid libraries for XML and HTTP would be a jolly good thing. As >you know, there is plenty of raw material around, but it obviously >doesn't do what you want, yet. I think I'm very close, based on the HXml Toolkit software and a few bits. I was parsing XML directly from the web earlier today, using Hugs on Win32. >If you, together with other interested parties, were to get together to >do this, the rest of us would be very grateful. You could do that by >directly working on the libraries distributed with GHC and Hugs (we >could give you CVS access). I'd like to see this become part of the standard libraries, when it becomes sufficiently stable and portable. I can't say how Uwe Schmidt feels about this. > Or you could make a separate project of it, >and distribute the stuff separately, which would free you from the GHC >and Hugs's release cycles. If the new library infrastructure project takes off, that could be an attractive option. I assume that in due course, the hierarchical libraries would move into such a framework. My goal is to be able to write software that uses an XML library that is easy for people not experienced with Haskell to obtain and load into their running environment. #g ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact

1 0

Weaving the Web with Haskell
by Graham Klyne 03 Feb '04

03 Feb '04

For the past week or so, I've been wrestling with various bits of Haskell libraries trying to get an XML parser running under Windows. Each time I think I've resolved a problem, another pops up to take its place. So I felt it was a good time to stand back and review my goals and approach. Background I believe that Haskell has a number of characteristics that make it eminently suitable for prototyping and deploying a range of web technologies. My own interest is in the Semantic Web area, but I think Haskell could also be usefully applied to Web Services and other web applications, and (as a language) potentially stands up well against the likes of Java and Python. The Web is spawning a host of little (and not-so-little) languages, and Haskell seems to have the right features to handle these in a clean, principled fashion. This is my working premise. Contemporary web applications are largely based on XML and HTTP. So a prerequisite for widespread use of any language for Web applications is good XML and HTTP support. Given the right libraries, Haskell could make handling these aspects very straightforward. But there seems to be a lack. Requirements Ideally, I think that the following requirements should be satisfied: - full XML parser supporting all features of the current XML core and namespace specifications. - full HTTP support, including content negotiation, redirection, read/write, GET, HEAD, PUT, POST, etc. - available across all Haskell compilers on all platforms Practically, I think the following would be enough to make a big difference, and this is what I'm currently looking for: - XML support, not necessary validating, but accepting full XML syntax and supporting including internal entity definitions and substitution and XML namespaces. - capability to access the representation of a Web Resource using HTTP GET - supported uniformly by GHC and Hugs, on Linux and Windows platforms. Where I am I have been trying to use the HXML toolbox, because I understand it's the only XML parser for Haskell that supports XML namespaces. Unfortunately, it seems to be rather dependent on older versions of GHC (unless I'm missing something), which is making it more problematic to adopt than I had hoped. Other packages, such as HaXml, seem to be more portable but as far as I'm aware are missing key functionality (notably XML namespace support). Looking forward I shall plug away for a little while with HXML toolbox -- currently I'm stuck for an MD5 library module -- and it may be that, with a little help, I'll win through: if I get it running with Hugs under Windows, I'll consider that to be significant progress. But maybe I'm missing a trick: is there other software I should be considering to satisfy my goals? Constructive suggestions would be welcome. I'm anticipating that as the library infrastructure project settles, we'll see some effort to move more of this kind of support code into a common and generally usable function library. I hope I shall be able to contribute to such an outcome. I suggest that the XML and HTTP support I've mentioned would be a key enabler to allow Haskell to be more widely used for real-world projects as well as academic research into language design issues (which currently seems to be its strongest developer community interest). #g ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact

5 4

RE: [Haskell] Weaving the Web with Haskell
by Simon Peyton-Jones 03 Feb '04

03 Feb '04

Graham Having solid libraries for XML and HTTP would be a jolly good thing. As you know, there is plenty of raw material around, but it obviously doesn't do what you want, yet. If you, together with other interested parties, were to get together to do this, the rest of us would be very grateful. You could do that by directly working on the libraries distributed with GHC and Hugs (we could give you CVS access). Or you could make a separate project of it, and distribute the stuff separately, which would free you from the GHC and Hugs's release cycles. Simon | -----Original Message----- | From: haskell-bounces(a)haskell.org [mailto:haskell-bounces@haskell.org] On Behalf Of Graham Klyne | Sent: 28 January 2004 12:37 | To: Haskell Mailing List | Subject: [Haskell] Weaving the Web with Haskell | | For the past week or so, I've been wrestling with various bits of Haskell | libraries trying to get an XML parser running under Windows. Each time I | think I've resolved a problem, another pops up to take its place. So I | felt it was a good time to stand back and review my goals and approach. | | | Background | | I believe that Haskell has a number of characteristics that make it | eminently suitable for prototyping and deploying a range of web | technologies. My own interest is in the Semantic Web area, but I think | Haskell could also be usefully applied to Web Services and other web | applications, and (as a language) potentially stands up well against the | likes of Java and Python. The Web is spawning a host of little (and | not-so-little) languages, and Haskell seems to have the right features to | handle these in a clean, principled fashion. This is my working premise. | | Contemporary web applications are largely based on XML and HTTP. So a | prerequisite for widespread use of any language for Web applications is | good XML and HTTP support. Given the right libraries, Haskell could make | handling these aspects very straightforward. But there seems to be a lack. | | | Requirements | | Ideally, I think that the following requirements should be satisfied: | - full XML parser supporting all features of the current XML core and | namespace specifications. | - full HTTP support, including content negotiation, redirection, | read/write, GET, HEAD, PUT, POST, etc. | - available across all Haskell compilers on all platforms | | Practically, I think the following would be enough to make a big | difference, and this is what I'm currently looking for: | - XML support, not necessary validating, but accepting full XML syntax and | supporting including internal entity definitions and substitution and XML | namespaces. | - capability to access the representation of a Web Resource using HTTP GET | - supported uniformly by GHC and Hugs, on Linux and Windows platforms. | | | Where I am | | I have been trying to use the HXML toolbox, because I understand it's the | only XML parser for Haskell that supports XML namespaces. Unfortunately, | it seems to be rather dependent on older versions of GHC (unless I'm | missing something), which is making it more problematic to adopt than I had | hoped. Other packages, such as HaXml, seem to be more portable but as far | as I'm aware are missing key functionality (notably XML namespace support). | | | Looking forward | | I shall plug away for a little while with HXML toolbox -- currently I'm | stuck for an MD5 library module -- and it may be that, with a little help, | I'll win through: if I get it running with Hugs under Windows, I'll | consider that to be significant progress. | | But maybe I'm missing a trick: is there other software I should be | considering to satisfy my goals? Constructive suggestions would be welcome. | | I'm anticipating that as the library infrastructure project settles, we'll | see some effort to move more of this kind of support code into a common and | generally usable function library. I hope I shall be able to contribute to | such an outcome. | | I suggest that the XML and HTTP support I've mentioned would be a key | enabler to allow Haskell to be more widely used for real-world projects as | well as academic research into language design issues (which currently | seems to be its strongest developer community interest). | | #g | | | ------------ | Graham Klyne | For email: | http://www.ninebynine.org/#Contact | | _______________________________________________ | Haskell mailing list | Haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/haskell

1 0

Spawning a process using Hugs/Win32
by Graham Klyne 03 Feb '04

03 Feb '04

I thought I might have a version of POpen that works with Win32/Hugs, but I'm only slightly successful. The function of POpen is to spawn a sub-process using the Win32 equivalent of pipes to supply standard input and to accept standard output and standard error from the spawned process. My starting point was a module called Win32Spawn, which I found in hslibs/win32 in the GHC source distribution. I have managed to compile the accompanying module and FFI interface into a DLL. The files used to do this are at [1], with the build process outlined in [2]. The key code in the new module POPen.hs [3] is in function "spawn", which was adapted from the original module "hslibs/win32/Win32Spawn.hs". I had to change the code for "spawn", because the Hugs interface for function openFd is has fewer parameters than that provided by GHC. The remaining parameters seem to be the same as for the GHC version of openFd. The resulting program has compiled successfully, and the spawned program is activated as requested (in this case, curl: my evidence is that my personal firewall is asking me if curl should be allowed to access the Internet). But I also get an I/O exception in the calling program, in function spawn, when trying to open a handle to write the spawned process' standard input stream. My evidence for this is by this modification to the code: [[ hIn <- catch (openFd (fromIntegral wIn) False WriteMode False) (\er -> do { putStrLn ("openFd wIn: "++show er) ; ioError er } ) ]] generating this message: [[ openFd wIn: openFd: does not exist (file does not exist) ]] My question is this: are there any fundamental differences between GHC and Hugs, either in the implementation of openFd, or in the FFI interface, that would cause (what I assume is) a divergence of behaviour? Does anyone have any other clues what might be going wrong here? (BTW, I did get the HXml Toolbox to read and parse XML *from the web* using the internal HTTP library.) #g -- [1] http://www.ninebynine.org/Software/HaskellUtils/Win32/ [2] http://www.ninebynine.org/Software/HaskellUtils/Win32/MakePOpenDLL.bat [3] http://www.ninebynine.org/Software/HaskellUtils/Win32/POpen.hs [4] http://www.ninebynine.org/Software/HaskellUtils/Win32/Win32Spawn.hs ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact

1 0

ANNOUNCE: HasFuse V1.1
by David Sabel 03 Feb '04

03 Feb '04

HasFuse Version 1.1 =================== HasFuse is a modification of the Glasgow Haskell Compiler which has built-in some precautions for the execution of unsafePerformIO, hence the name "HasFuse". This means the user does not have to undertake the proof obligation that the timing of an I/O-operation wrapped by unsafePerformIO is independent from each other I/O-operation. HasFuse is available as a patch of the GHC 5.04.3 source code from http://www.ki.informatik.uni-frankfurt.de/~sabel/hasfuse On the same location, building instructions and further documentation can be found. The semantics of programs which make use of unsafePerformIO is given by the nondeterministic call-by-need calculus FUNDIO. This calculus provides an I/O-interface, which can be used to model direct-call I/O within Haskell. The FUNDIO-calculus is published in the technical report "Frank-16", see http://www.ki.informatik.uni-frankfurt.de/papers/articles.html#Frank-16 The results of applying the FUNDIO semantics to Haskell and the (FUNDIO-) correctness of those program transformations which are performed by the GHC are available as "Technical Report Frank-17" through http://www.ki.informatik.uni-frankfurt.de/papers/articles.html#Frank-17 Changes from version 1.0 to 1.1: -------------------------------- HasFuse provides a new flag -fno-hasfuse, which runs the GHC in GHC mode, i.e. all optimisations performed by the original GHC are turned on.

1 0

ANNOUNCE: HSQL 1.2 release
by Krasimir Angelov 02 Feb '04

02 Feb '04

Dear Haskellers, I am pleased to announce the HSQL library version 1.2. The HSQL is a simple library, which provides interface to multiple databases. MySQL, PostgreSQL, ODBC and SQLite are currently supported. What is new? ~~~~~~~~~~~~ 1. The library namespaces are changed as follows: - The namespace for ODBC is Database.HSQL.ODBC instead of Database.ODBC.HSQL - The namespace for PostgreSQL is Database.HSQL.PostgreSQL instead of Database.PostgreSQL.HSQL - The new namespace for MySQL is Database.HSQL.MySQL 2. A new Database.HSQL module, which is an abstract database interface, is added. Each database specific module exports only the "connect" function, which returns the abstract Connection type. All functions, which operate on the Connection are exported from the Database.HSQL. The user can import the database specific module only in place where the connection is established and for all other places it is enough to import only Database.HSQL. The abstraction simplifies the development of database independent applications and libraries because the dependent part will be only in place where the connection is established. 3. A module for SQLite database (Database.HSQL.SQLite) has been added. The SQLite is a simple database library, which works directly with a file and doesn't use database server. 4. The new version works not only with GHC but also with Hugs under both Windows and Linux. 5. A simple interface to the database meta information has been added. By using the interface the application can determine the list of tables and views in the database and the list of columns in each table. 6. Some bugfixes and optimisations. Great thanks to Bjorn Bringert for his work on porting of HSQL to Hugs and on the meta information API. Home Page ~~~~~~~~~ http://htoolkit.sourceforge.net Cheers, Krasimir __________________________________ Do you Yahoo!? Yahoo! SiteBuilder - Free web site building tool. Try it! http://webhosting.yahoo.com/ps/sb/

1 0