Re: [Haskell-cafe] parsec2 vs. parsec3... again
> This isn't completely without basis. For instance, I made some big speed > improvements to attoparsec's very performance-sensitive takeWhile function > just the other day, thanks to -auto-all. > > I might, though, see if there's a way I could enable that flag only for > myself (in a way that I wouldn't routinely forget). See ~/.cabal/config - I use that to make sure all my packages are installed globally with profiling, and I think it might have enough options to force -auto-all in some way. Thanks, Neil ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] hoogle command line program on Linux
On Mon, Jan 10, 2011 at 10:17 PM, Neil Mitchell wrote: >>> If so, I'll make a new release that just changes the >>> file creation mask to the above during hoogle data (and sets it back >>> after). Thanks to Erik's help testing preview versions I've now released Hoogle 4.1.4 that sets the file creation mask appropriately. Thanks, Neil ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] hoogle command line program on Linux
On 15 January 2011 22:53, Neil Mitchell wrote: > On Mon, Jan 10, 2011 at 10:17 PM, Neil Mitchell wrote: If so, I'll make a new release that just changes the file creation mask to the above during hoogle data (and sets it back after). > > Thanks to Erik's help testing preview versions I've now released > Hoogle 4.1.4 that sets the file creation mask appropriately. Shouldn't data like this really go in /var rather than /usr ? To quote Wikipedia [1]: "/var/: Variable files—files whose content is expected to continually change during normal operation of the system—such as logs, spool files, and temporary e-mail files." [1]: http://en.wikipedia.org/wiki/Filesystem_Hierarchy_Standard -- Ivan Lazar Miljenovic ivan.miljeno...@gmail.com IvanMiljenovic.wordpress.com ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] hoogle command line program on Linux
>> Thanks to Erik's help testing preview versions I've now released >> Hoogle 4.1.4 that sets the file creation mask appropriately. > > Shouldn't data like this really go in /var rather than /usr ? To > quote Wikipedia [1]: "/var/: Variable files—files whose content is > expected to continually change during normal operation of the > system—such as logs, spool files, and temporary e-mail files." The Hoogle databases are expected to change very rarely - most users will install them when they install Hoogle. A small number will update them occasionally as the packages update. I'm using the Cabal datadir to store the databases, but does Cabal provide a more sensible place to put them? As a Windows user, I'm happy to defer to what Posix system users want. Thanks, Neil ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] hoogle command line program on Linux
Hi, Am Samstag, den 15.01.2011, 13:38 + schrieb Neil Mitchell: > >> Thanks to Erik's help testing preview versions I've now released > >> Hoogle 4.1.4 that sets the file creation mask appropriately. > > > > Shouldn't data like this really go in /var rather than /usr ? To > > quote Wikipedia [1]: "/var/: Variable files—files whose content is > > expected to continually change during normal operation of the > > system—such as logs, spool files, and temporary e-mail files." > > The Hoogle databases are expected to change very rarely - most users > will install them when they install Hoogle. A small number will update > them occasionally as the packages update. I'm using the Cabal datadir > to store the databases, but does Cabal provide a more sensible place > to put them? As a Windows user, I'm happy to defer to what Posix > system users want. shouldn’t they change with every library update? In that case, maybe that should be managed by cabal, similar to how cabal can update your haddock index at ~/.cabal/share/doc/index.html. (Such an infrastructure would be nice for other people as well, e.g. leksah). But I might misunderstand how hoogle works and what the databases are for. Greetings, Joachim -- Joachim "nomeata" Breitner mail: m...@joachim-breitner.de | ICQ# 74513189 | GPG-Key: 4743206C JID: nome...@joachim-breitner.de | http://www.joachim-breitner.de/ Debian Developer: nome...@debian.org signature.asc Description: This is a digitally signed message part ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Needed: A repeatable process for installing GHC on Windows
>> Earlier today I was trying to set up a Windows build bot for the >> 'network' package. That turned out to be quite difficult. Too much >> playing with PATHs, different gcc versions, etc. Does anyone have a >> repeatable, step-by-step process to install GHC and get a build >> environment (where I could build network) going? No, but I used to (and sadly can't find it any more). I used to have a script called ghcsetup which built GHC on Windows, and importantly validated the setup was correct (the right gcc was first in the path etc) and took actions to correct it. I am sure there used to be a great web page on the GHC wiki, saying the exact steps to build (written by Claus), but I can't find it any more. Perhaps Claus knows where it has gone? Thanks, Neil ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] hoogle command line program on Linux
Hi Joachim, >> The Hoogle databases are expected to change very rarely - most users >> will install them when they install Hoogle. A small number will update >> them occasionally as the packages update. I'm using the Cabal datadir >> to store the databases, but does Cabal provide a more sensible place >> to put them? As a Windows user, I'm happy to defer to what Posix >> system users want. > > shouldn’t they change with every library update? In that case, maybe > that should be managed by cabal, similar to how cabal can update your > haddock index at ~/.cabal/share/doc/index.html. Yes, that's exactly how it should work. All we need is someone to do the work :-) There is a bug tracking the progress on this issue here: http://code.google.com/p/ndmitchell/issues/detail?id=80 - but its a reasonable amount of work, touching lots of different parts of the Haskell infra. Thanks, Neil ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Needed: A repeatable process for installing GHC on Windows
Earlier today I was trying to set up a Windows build bot for the 'network' package. That turned out to be quite difficult. Too much playing with PATHs, different gcc versions, etc. Does anyone have a repeatable, step-by-step process to install GHC and get a build environment (where I could build network) going? If you don't need to build GHC yourself, then the binary installers (or daily snapshot builds) should come with their own copy of GCC tools. So you only need some shell/autoconf environment on top (either cygwin or MSYS). Fewer things to go wrong than in the past, but cygwin's configure scripts tend to think they're on unix, and for MSYS, it can be difficult to get the right pieces together. Snapshots should appear as installers and tar-balls here: http://www.haskell.org/ghc/download#snapshots but windows build failures tend to go unnoticed for a while, so you might have to send a heads-up to cvs-ghc@.. if you need up to date builds. No, but I used to (and sadly can't find it any more). I used to have a script called ghcsetup which built GHC on Windows, and importantly validated the setup was correct (the right gcc was first in the path etc) and took actions to correct it. I am sure there used to be a great web page on the GHC wiki, saying the exact steps to build (written by Claus), but I can't find it any more. Perhaps Claus knows where it has gone? Yes, I ran into this so often that I made a record of the "build your own GHC" steps one time, together with a cygwin package description that recorded the cygwin packages needed for such builds. Both cygwin and GHC have moved since then, and I guess someone decided it was too much trouble to keep the step-by-step guide up to date, or perhaps it was no longer needed (check with GHC HQ to be sure)? If you do need to build your own GHC head, the GHC wiki building guide has a page on "Setting up a Windows system for building GHC" http://hackage.haskell.org/trac/ghc/wiki/Building/Preparation/Windows which not only claims to list suitable MSYS versions, but holds copies (so you don't have to hunt for the MSYS packages), as well as links to other necessary tools (python, alex, happy, darcs). What I do not know is whether the versions are still up to date (check with GHC HQ). As usual, if the wiki page instructions are not working anymore, please add a note, and include any improvements you figure out;-) Claus PS. GHC no longer uses buildbot, it has its own builder, so if you really want to set up a build bot, you can probably copy GHC HQ's setup for the purpose? ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Needed: A repeatable process for installing GHC on Windows
I've managed to set up a working MinGW/MSYS environment, build the network package, and document the steps here: http://blog.johantibell.com/2011/01/setting-up-haskell-development.html Johan ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] ANN: extcore 1.0
On Thu, Jan 13, 2011 at 11:45 AM, Tim Chevalier wrote: > Hello, > > I've recently released version 1.0 of extcore, a library for > processing code in GHC's text-based External Core format. extcore > includes a parser, prettyprinter, typechecker, and interpreter for > External Core, as well as modules for computing module dependencies > and combining multiple Core modules into a single module. The library > was originally written by Andrew Tolmach and was part of the GHC > distribution. After becoming its maintainer, about a year ago I moved > the library to Hackage. Since in the past year, the library has > acquired some users but has received no fatal bug reports, I thought > it was time for a 1.0 release. > > You might be interested in using this library if you're a compiler or > tool writer who wants to write a tool that processes the output of > GHC's front-end and simplifier -- that is, code in the Core > intermediate language, a typed functional language in which many of > Haskell's features have been desugared into a polymorphically typed > lambda-calculus (extended with type coercions, as are necessary to > support type system features such as GADTs). Such tools might include > alternative back-ends or new program transformations. Using the > extcore library allows tool writers to experiment with implementing > new compiler features without modifying GHC itself. While the GHC API > is also a route to adding new stages to the GHC pipeline, using > extcore has the advantage that External Core is well-specified (read > on) and less subject to change than GHC's internal Core format. The > library has been tested on External Core produced by GHC 6.10.* and > GHC 6.12.*. It has not yet been tested on External Core produced by > GHC 7.*. > > Documentation for the External Core format itself lives at > http://www.haskell.org/ghc/docs/latest/html/ext-core/core.pdf - but Oops, that'll learn me not to check links before I send email. The correct URL for the documentation is: http://www.haskell.org/ghc/docs/latest/core.pdf > documentation for the library is limited, and I would welcome > documentation or other patches. Currently the best source of > documentation for the library is the README file, found in the > distribution. To get the distribution, please visit > http://hackage.haskell.org/package/extcore or cabal install extcore. > > Please direct replies to glasgow-haskell-us...@haskell.org, with me > CCed. Please make sure not to reply to hask...@haskell.org. > Cheers, Tim -- Tim Chevalier * http://cs.pdx.edu/~tjc/ * Often in error, never in doubt "an intelligent person fights for lost causes,realizing that others are merely effects" -- E.E. Cummings ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Guy Steele's Praise For Haskell @ Strange Loop Keynote
Guy Steele did the keynote on parallelism [1] at the Strange Loop [2] conference in which he said that he could do it over Fortress [3] would have been modeled on Haskell rather than Fortran. The relevant portions are between 49:36 - 49:50. Thought it might interest readers of this list. -deech [1] http://www.infoq.com/presentations/Thinking-Parallel-Programming [2] http://strangeloop2010.com/ [3] http://en.wikipedia.org/wiki/Fortress_(programming_language) ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Guy Steele's Praise For Haskell @ Strange Loop Keynote
Pretty interesting links, thanks. Unfortunately, if Fortress is to have any chance of success with programmers, it will need to be straight-line and essentially have Algol-based syntax. MATLAB, LabVIEW, Fortran, Java, C, and non-OO C++/random subsets of C++ rule scientific programming. Unit testing is rare and sporadic. In dragging scientists halfway to something new, the exotic, powerful things in Haskell will have to be left behind, just as Java only has a tiny fraction of what Smalltalk has had since the '80s. That seems clear to me, anyway. Warren On Sat, Jan 15, 2011 at 2:31 PM, aditya siram wrote: > Guy Steele did the keynote on parallelism [1] at the Strange Loop [2] > conference in which he said that he could do it over Fortress [3] > would have been modeled on Haskell rather than Fortran. The relevant > portions are between 49:36 - 49:50. Thought it might interest readers > of this list. > > -deech > > [1] http://www.infoq.com/presentations/Thinking-Parallel-Programming > [2] http://strangeloop2010.com/ > [3] http://en.wikipedia.org/wiki/Fortress_(programming_language) ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Guy Steele's Praise For Haskell @ Strange Loop Keynote
So everybody doesn't have to go watch it, here is a shortened version of what Steele said in the video: Although Fortress is originally designed as an object-oriented framework in which to build an array-style scientific programming language, [...] as we've experimented with it and tried to get the parallelism going we found ourselves pushed more and more in the direction of using immutable data structures and a functional style of programming. [...] If I'd known seven years ago what I know now, I would have started with Haskell and pushed it a tenth of the way toward Fortran instead of starting with Fortran and pushing it nine tenths of the way toward Haskell. I think I might use this in some slides soon. :) Thanks for pointing it out! - Jake ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Misleading MVar documentation
On Wed, Jan 12, 2011 at 11:23 AM, Neil Brown wrote: > On 12/01/11 15:53, Edward Z. Yang wrote: >> >> These are interesting, opposed perspectives, and I suspect what would be >> good is to treat both situations. I think perhaps what would be good >> to put in the introduction is the conceptual model of MVars: that is, >> take and put are the fundamental operations, and everything else is >> composed of them. With additional constraints on who is writing and >> reading >> MVars, you can assume more safety properties, but you have to ensure >> that those are indeed held (or you should use STM instead.) >> >> I'll try another writeup. Does anyone know where the original papers for >> MVars might be? >> > I think the original paper is "Concurrent Haskell", available here: > > http://www.haskell.org/ghc/docs/papers/concurrent-haskell.ps.gz > > and here: > > http://research.microsoft.com/en-us/um/people/simonpj/papers/concurrent-haskell.ps.gz Actually, the first presentation of M-structures is rather older than that. See Barth, Nikhil, and Arvind's FPCA '91 paper: http://portal.acm.org/citation.cfm?id=652538 The original formulation was indeed in terms of "take" and "put", though unconditional read and write primitives were prtty commonly used in Id programs. The take/put view can also usefully be thought of as a 1-element blocking channel. -Jan-Willem MAessen > > Thanks, > > Neil. > > ___ > Haskell-Cafe mailing list > Haskell-Cafe@haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe > ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Guy Steele's Praise For Haskell @ Strange Loop Keynote
On Sat, Jan 15, 2011 at 9:02 PM, Jake McArthur wrote: > So everybody doesn't have to go watch it, here is a shortened version of > what Steele said in the video: > >> Although Fortress is originally designed as an object-oriented framework >> in which to build an array-style scientific programming language, [...] as >> we've experimented with it and tried to get the parallelism going we found >> ourselves pushed more and more in the direction of using immutable data >> structures and a functional style of programming. [...] If I'd known seven >> years ago what I know now, I would have started with Haskell and pushed it a >> tenth of the way toward Fortran instead of starting with Fortran and pushing >> it nine tenths of the way toward Haskell. > > I think I might use this in some slides soon. :) Thanks for pointing it out! The big things I can recall missing were pattern matching and Haskell-style classes rather than OO + generic typing. The Fortress type system actually approximates pattern matching in some interesting ways, but it's not the same. -Jan-Willem Maessen Experienced Fortress programmer (!) > > - Jake > > ___ > Haskell-Cafe mailing list > Haskell-Cafe@haskell.org > http://www.haskell.org/mailman/listinfo/haskell-cafe > ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Misleading MVar documentation
> I'll try another writeup. Does anyone know where the original papers for > MVars might be? In my mind, full/empty variables date back to dataflow machines. But I don't know where I got this idea. Stefan ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Guy Steele's Praise For Haskell @ Strange Loop Keynote
> > > >> Although Fortress is originally designed as an object-oriented framework > >> in which to build an array-style scientific programming language, [...] > as > >> we've experimented with it and tried to get the parallelism going we > found > >> ourselves pushed more and more in the direction of using immutable data > >> structures and a functional style of programming. [...] If I'd known > seven > >> years ago what I know now, I would have started with Haskell and pushed > it a > >> tenth of the way toward Fortran instead of starting with Fortran and > pushing > >> it nine tenths of the way toward Haskell. > This is at 49th minute into the talk :) Really nice to hear him say this. Regards, Kashyap ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe