read the QML paper! that's the trick. simpler, but with a heavier annotation burden than the more sophisticated approaches | -----Original Message----- | From: Gregory Crosswhite [mailto:gcross@phys.washington.edu] | Sent: 19 November 2010 19:59 | To: Simon Peyton-Jones | Cc: Haskell Cafe; glasgow-haskell-users@haskell.org | Subject: Re: [Haskell-cafe] Impredicative Types | | Awesome, thank you. :-) One more question out of pure curiosity, if | you have the time: What is allowing you to keep them in? I thought | that the problem was that they made a mess that touched every party of | the type checker rather than being centralized in one place. Was there | a trick you discovered that now allows you to support them without | creating such a mess? | | Cheers, | Greg | | On 11/19/10 11:45 AM, Simon Peyton-Jones wrote: | > Yes, impredicative types are still in, but in a simpler form than before, along the | lines of QML | > http://research.microsoft.com/en-us/um/people/crusso/qml/ | > | > I have been too busy with getting the new type checker working to describe or | document it. Notably, I have not yet added syntax for QML's rigid type annotations, | which leads to a loss of expressive power. | > | > too much to do! | > | > Bottom line: if you are interested in impredicative polymorphism, let me know. A | good way to do so would be to register your interest on | > http://hackage.haskell.org/trac/ghc/ticket/4295 | > Just add your email address to the cc list, *and* write a comment giving an example | of how you are using impredicative poly, and pointing to any further info. That'll help | motivate me to do the remaining work! | > | > Thanks | > | > Simon | > | > | -----Original Message----- | > | From: haskell-cafe-bounces@haskell.org [mailto:haskell-cafe- | bounces@haskell.org] | > | On Behalf Of Gregory Crosswhite | > | Sent: 19 November 2010 19:23 | > | To: Haskell Cafe | > | Subject: [Haskell-cafe] Impredicative Types | > | | > | Hey everyone! I haven't had a chance to try out GHC 7 myself, but I saw | > | in the documentation that Impredicative Types are still supported. Is | > | this true? I thought that they were on their way out because they | > | overcomplicated type checking; has this plan been changed? | > | | > | Cheers, | > | Greg | > | _______________________________________________ | > | Haskell-Cafe mailing list | > | Haskell-Cafe@haskell.org | > | http://www.haskell.org/mailman/listinfo/haskell-cafe | > |

On 11/19/10 12:22 PM, Andrew Coppin wrote:

Use the Force I tried, but I'm not yet strong enough in the Force to read citations with my eyes closed. :-(

(PS: Thanks for pointing me to the paper, Simon!) Cheers, Greg