[Haskell-cafe] Interest in helping w/ Haskell standard
At the end of the Haskell Workshop at ICFP, we had the traditional "Future of Haskell" discussion (chaired by Andres Loeh). One of the main topics was the perceived need of a new standard, because the Haskell 98 standard is quite old already, and Haskell has evolved in the meantime, leading to a situation where almost none of the actually existing Haskell programs is according to the 98 standard. No clear opinion was visible on what form a new standard would take. There was, however, considerable support for the idea to standardize an incremental and moderate extension to Haskell 98 (working name "Haskell 06" or "Industrial Haskell"). This effort would then be separate from discussion about the Real Next Version (dubbed "Haskell 2"). John Launchbury asked for a show of hands of those who would be interested in helping out with the "Haskell 06" standard. I think "helping" means "willing to spend a non-trivial amount of time". That is, it's pretty well expected that most Haskellers will be willing to contribute to discussion on the mailing lists, but we're trying to get a list of those who want to take it to the next level. If you raised your hand, or if you think this describes you, please email John Launchbury at [EMAIL PROTECTED] peace, Isaac Jones & Andres Loeh ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Interest in helping w/ Haskell standard
On 10/13/05, Isaac Jones <[EMAIL PROTECTED]> wrote: > At the end of the Haskell Workshop at ICFP, we had the traditional > "Future of Haskell" discussion (chaired by Andres Loeh). One of the > main topics was the perceived need of a new standard, because the > Haskell 98 standard is quite old already, and Haskell has evolved in > the meantime, leading to a situation where almost none of the actually > existing Haskell programs is according to the 98 standard. No clear > opinion was visible on what form a new standard would take. There was, > however, considerable support for the idea to standardize an > incremental and moderate extension to Haskell 98 (working name > "Haskell 06" or "Industrial Haskell"). This effort would then be > separate from discussion about the Real Next Version (dubbed "Haskell > 2"). > > John Launchbury asked for a show of hands of those who would be > interested in helping out with the "Haskell 06" standard. I think > "helping" means "willing to spend a non-trivial amount of time". That > is, it's pretty well expected that most Haskellers will be willing to > contribute to discussion on the mailing lists, but we're trying to get > a list of those who want to take it to the next level. If you raised > your hand, or if you think this describes you, please email John > Launchbury at [EMAIL PROTECTED] > I'm wondering what "incremental and moderate" extension means? Does it mean "completely backwards compatible" or can it mean completely new features including ones which subsume existing ones (I'm specifically interested in seeing SPJ's records proposal included, and a new module system). I'd also like to point out the haskell wiki page: http://www.haskell.org/hawiki/HaskellOhSix Which already contain some desired features. Perhaps a running summary of "definate", "maybe", and "not until Haskell 2" features could be kept there as discussions progress? /S -- Sebastian Sylvan +46(0)736-818655 UIN: 44640862 ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Interest in helping w/ Haskell standard
Sebastian Sylvan wrote: > I'm wondering what "incremental and moderate" extension means? I don't know what others mean by it, but for me, it implies standardizing existing practice, with possibly some conservative redesign to get rid of any hysterical warts. This is, BTW, what the C89 standard did for C, and it was a highly successful standard. -- Antti-Juhani ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
RE: [Haskell-cafe] Interest in helping w/ Haskell standard
On 12 October 2005 23:50, Sebastian Sylvan wrote: > (I'm specifically interested in seeing SPJ's records proposal > included, and a new module system). Highly unlikely, IMHO. A new revision of the Haskell standard is not the place for testing new research, rather it's a clear specification of existing well-understood language features. If you want a new record system, or a new module system, now is the time to start designing and implementing them ready for the next standardisation process. Cheers, Simon ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Interest in helping w/ Haskell standard
On Thursday 13 October 2005 09:42, Simon Marlow wrote: > On 12 October 2005 23:50, Sebastian Sylvan wrote: > > (I'm specifically interested in seeing SPJ's records proposal > > included, and a new module system). > > Highly unlikely, IMHO. A new revision of the Haskell standard is not > the place for testing new research, rather it's a clear specification of > existing well-understood language features. ... and, in the case of the Standard Prelude section, or equivalent, a specification of well-understood functions that the spec authors agree "should" be provided in all implementations. I would hope that arbitrary binary I/O - to cite an important example - would not be considered a "research topic", as it is actually quite trivial. (It is, however, one of the "must haves" for Haskell to be considered for use in production systems.) -- Robin ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Interest in helping w/ Haskell standard
On 10/13/05, Simon Marlow <[EMAIL PROTECTED]> wrote: > On 12 October 2005 23:50, Sebastian Sylvan wrote: > > > (I'm specifically interested in seeing SPJ's records proposal > > included, and a new module system). > > Highly unlikely, IMHO. A new revision of the Haskell standard is not > the place for testing new research, rather it's a clear specification of > existing well-understood language features. > I can certainly understand this point of view. I am (as primarily a user and not a language designer) perhaps too eager to get my hands on cool new stuff :-) > If you want a new record system, or a new module system, now is the time > to start designing and implementing them ready for the next > standardisation process. Okay then. Consider this my contribution to the discussion. First of all I would like to urge the people who do end up working on this to seriously consider replacing H98's records system. I may be wrong but the impression I get is that enough people dislike the current system enough to warrant a replacement. And to me it seems to be a pretty much "slam-dunk" case that the proposal is *a lot* better than what we current have. On the module system. You may consider this a proposal. There have been discussions on this mailing lists about it but let me recap the main gist of it. Conservatively extend the current ghc hierarchical module system by allowing you to re-export modules "qualified". So you could write module GTK (..., qualified module GTK.Button as Button, ...) where Then the user could just import GTK and get all of the contents of GTK.Button imported qualifed as Button automatically. This allow libraries written using the current hierarchcical system to work without any changes, while new libraries can use the new feature (I'm guessing GTK2HS would benefit a lot from this, getting rid of all the ugly "buttonNew"-style functions). /S -- Sebastian Sylvan +46(0)736-818655 UIN: 44640862 ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
RE: [Haskell-cafe] Interest in helping w/ Haskell standard
On Thu, 13 Oct 2005, Simon Marlow wrote: On 12 October 2005 23:50, Sebastian Sylvan wrote: (I'm specifically interested in seeing SPJ's records proposal included, and a new module system). Highly unlikely, IMHO. A new revision of the Haskell standard is not the place for testing new research, rather it's a clear specification of existing well-understood language features. In that context, how well-understood is the combination of impredicative types via boxy types and a proper existential quantifier at the moment? It's certainly something that has many uses in an industrial context. -- [EMAIL PROTECTED] Society does not owe people jobs. Society owes it to itself to find people jobs. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Interest in helping w/ Haskell standard
On Thu, Oct 13, 2005 at 10:42:24AM +0100, Simon Marlow wrote: > On 12 October 2005 23:50, Sebastian Sylvan wrote: > > (I'm specifically interested in seeing SPJ's records proposal > > included, and a new module system). > > Highly unlikely, IMHO. A new revision of the Haskell standard is not > the place for testing new research, rather it's a clear specification of > existing well-understood language features. Indeed. I'd love to have a new Haskell 06 that adds just a carefully (and conservatively) selected subset of ghc's extensions, if that meant that my code would also run on hugs and jhc. Like everyone else, there are *some* experimental features I'd love to see go in, most particularly John Meacham's class aliases proposal, which seems like (perhaps from my naive perspective) a moderately conservative change that opens up huge possibilities for improved libraries (which could then be developed to the new standard). -- David Roundy ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re[2]: [Haskell-cafe] Interest in helping w/ Haskell standard
Hello Sebastian, Thursday, October 13, 2005, 2:49:46 AM, you wrote: SS> I'm wondering what "incremental and moderate" extension means? SS> Does it mean "completely backwards compatible" or can it mean SS> completely new features including ones which subsume existing ones SS> (I'm specifically interested in seeing SPJ's records proposal SS> included, and a new module system). i must figure out that these extensions are not really implemented at this moment. imho, before including something in standard we need to play with it some time imho, features for new Haskell standard must include extensions common for Hugs and GHC (because implementing in both compilers proves their usefulness) and most wanted features from GHC. Haskell2 wiki page are close to list all these features -- Best regards, Bulatmailto:[EMAIL PROTECTED] ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re[2]: [Haskell-cafe] Interest in helping w/ Haskell standard
Hello Sebastian, Thursday, October 13, 2005, 4:09:55 PM, you wrote: >> > (I'm specifically interested in seeing SPJ's records proposal >> > included, and a new module system). SS> First of all I would like to urge the people who do end up working on SS> this to seriously consider replacing H98's records system. I may be yes, it is a common viewpoint, afaik. the only problem is what this new record system was never really implemented, partially because it is not backward-compatible with H98, partially because Simons are not very like such syntax sugar extensions, they prefer "real semantic beasts" :) so, this proposal is a bit out of luck :))) -- Best regards, Bulatmailto:[EMAIL PROTECTED] ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re[2]: [Haskell-cafe] Interest in helping w/ Haskell standard
Hello Simon, Thursday, October 13, 2005, 1:42:24 PM, you wrote: >> (I'm specifically interested in seeing SPJ's records proposal >> included, and a new module system). SM> Highly unlikely, IMHO. A new revision of the Haskell standard is not SM> the place for testing new research, rather it's a clear specification of SM> existing well-understood language features. +1 :) but we may add them to standard definition as "Possible extensions" appendix. moreover, if some already implemented language extensions (f.e., Template Haskell) are not good enough to be included in standard itself, it may be also included here - just to have one place where many language extensions are described. and also a source for new features which MAY be included in Haskell2 -- Best regards, Bulatmailto:[EMAIL PROTECTED] ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell] Re: [Haskell-cafe] Interest in helping w/ Haskell standard
(Trimming CC list. Maybe we should take this to haskell-cafe?) Sebastian Sylvan <[EMAIL PROTECTED]> writes: (snip quotes) > I'm wondering what "incremental and moderate" extension means? > Does it mean "completely backwards compatible" or can it mean > completely new features including ones which subsume existing ones > (I'm specifically interested in seeing SPJ's records proposal > included, and a new module system). I was intentionally not addressing that question, because it's pretty much The Question. I certainly don't know the answer; just trying to figure out who wants to get involved, as a first step. I think everyone is agreed, though, that any process is going to be a very open one. peace, isaac ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell] Re: [Haskell-cafe] Interest in helping w/ Haskell standard
On Thu, Oct 13, 2005 at 02:09:55PM +0200, Sebastian Sylvan wrote: > Okay then. Consider this my contribution to the discussion. > First of all I would like to urge the people who do end up working on > this to seriously consider replacing H98's records system. I may be > wrong but the impression I get is that enough people dislike the > current system enough to warrant a replacement. And to me it seems to > be a pretty much "slam-dunk" case that the proposal is *a lot* better > than what we current have. How about a standard HList-style library specified in the report? then we get fancy records without having to change the language. Then we could work on syntatic sugar to make HList records as easy to use as built in ones. John -- John Meacham - ⑆repetae.net⑆john⑈ ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
