RE: Local evidence and type class instances
| Now that the Glorious New type checker can handle local evidence | seamlessly, is it a big implementation burden to extend it to deal | with local *type class instances* in addition to local *equality | constraints*? ... | Are there big theoretical problems with this extension, or is it just | a lack of engineering effort that has prevented its implementation? The issues I know about are a) overlap (of course) b) coherence (two different instances in two places) c) if you add associated types, then soundness too d) completeness Concerning (c) you can't define (type instance F Int = Bool) in one place and (type instance F Int = Char) somewhere else, or you'll have seg faults. GHC checks that you don't, but it can only do that because the instances are global. Perhaps most important Concerning (d) we only have a theory for local *constraints*, not for local *constraint schemes* (ie with foralls in them). To get complete inference with assumptions like (F (G x) ~ G x) was tricky enough. Adding schemes would make it harder. But yes, it'd be useful. As well as Oleg and Ken's paper, the "Derivable type classes" paper that Ralf and I wrote some time ago argues for local constraint schemes. Simon ___ Glasgow-haskell-users mailing list Glasgow-haskell-users@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users
RE: ghc7 typing problem
Thanks. We knew about this -- in flight changes happening now -- but I've added it as a new regression test even so. Simon | -Original Message- | From: Christian Maeder [mailto:christian.mae...@dfki.de] | Sent: 15 October 2010 17:53 | To: GHC Users Mailing List | Cc: Simon Peyton-Jones | Subject: ghc7 typing problem | | Hi, | | I've looked at: | http://hackage.haskell.org/trac/ghc/blog/LetGeneralisationInGhc7 | | but still have a problem with the attached program that fails with: | | GHCi, version 7.1.20101010: http://www.haskell.org/ghc/ :? for help Loading package | ghc-prim ... linking ... done. | Loading package integer-gmp ... linking ... done. | Loading package base ... linking ... done. | Loading package ffi-1.0 ... linking ... done. | [1 of 1] Compiling Main ( HasKey.hs, interpreted ) | | HasKey.hs:17:10: | Couldn't match type `key' with `key1' | because this skolem type variable would escape: `key1' | This skolem is bound by the instance declaration | In the instance declaration for `Ord (Keyed x)' | | at the final instance for Ord. | | The original code comes from | http://hackage.haskell.org/packages/archive/uni-util/2.2.0.0/doc/html/Util- | VariableSet.html | | Cheers Christian ___ Glasgow-haskell-users mailing list Glasgow-haskell-users@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users
Re: ANNOUNCE: GHC 7.0.1 Release Candidate 1
>> I created >> http://hackage.haskell.org/trac/ghc/ticket/4354 and attached the >> patch. > > We now have a mixture of > org.haskell.ghc > and > org.haskell.GHC > > Is that deliberate, or should I lowercase them all? Not deliberate. I updated the patch in my client to lowercase them all. Haven't uploaded it yet because of the question below. >> The main question is whether or not I can kill Uninstaller, as it's >> even more broken now than it was before. If yes, I'll attach another >> patch that removes it. > > I'd prefer to fix it, if possible. Is there anything magic about the > name, or could we have Uninstaller-7.0.1 etc? I don't think there's anything magic about the name. The tricky thing about uninstallation is detecting when you uninstalled the currently "active" version (as defined by the plain 'ghc' symlinks and thy Current symlink), and either remove the links in that case, or pick the highest remaining version and relink it. And then there's the question of the man pages. We should probably install versioned man pages and play the same game with links for what happens when you type 'man ghc' (hopefully it's the version that runs when you type 'ghc'!). I'd also want to hook into the configure system to create a versioned uninstall script from a template... I assume there's already machinery for that stuff, I just don't know how to plug it in. If it's ok, I could submit the patch with the template version and let you plug in the magic to make configure replace the VERSION macro. This is all getting a little complicated. Well, it's not complicated for a real programming language, but this is shell we're talking about. So as long as I'm rewriting it, I'm tempted to do so in python, which all OS X installs have. So the steps would be: - install versioned man pages - add man pages to create-links - write a new uninstaller that does all the magic above WRT stuff like versioned man pages, I'm also reluctant to diverge from what the standard unix install does. How does it handle multiple versions and uninstalls? ___ Glasgow-haskell-users mailing list Glasgow-haskell-users@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users
Template Haskell: onward and upward
Folks Following lots of feedback from users, especially at ICFP, I've evolved some proposals for Template Haskell, that should make it both more expressive, and more secure. http://hackage.haskell.org/trac/ghc/blog/Template%20Haskell%20Proposal Do let me know what you think. Discussion by email is fine (cc me if it's on Haskell-cafe), or comments direct on the Trac. (None of this will be in GHC 7.0; it's a proposed plan for post-release improvements.) Thanks Simon ___ Glasgow-haskell-users mailing list Glasgow-haskell-users@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users
Re: Has something about -prof -auto-all changed in GHC 7?
On Tue, Oct 12, 2010 at 2:08 AM, Simon Marlow wrote: > Nothing has changed in this area that I'm aware of. Could you give more > details? > Sure. I'm building darcs HEAD of the text library using both 6.12.3 and 7: darcs get http://code.haskell.org/text cd text cabal install --enable-library-profiling It gets built with -prof -auto-all (see ghc-prof-opts in text.cabal). If you then go into the tests/benchmarks directory, you can build and run the Equality benchmark: ghc -prof -auto-all -O --make Equality (add -rtsopts if using 7) Grab this sample file: http://www.facebook.com/jobs_puzzles/twl06.txt Then run Equality on it: time ./Equality text twl06.txt +RTS -P I've attached samples of the .prof files generated by each version of GHC so you can see the rather dramatic difference. To my eye, it looks like something is quite wrong with the version from 7. Mon Oct 18 21:38 2010 Time and Allocation Profiling Report (Final) Equality +RTS -P -RTS text /home/bos/Download/twl06.txt total time = 10.76 secs (538 ticks @ 20 ms) total alloc = 8,639,840,712 bytes (excludes profiling overheads) COST CENTREMODULE %time %alloc ticks bytes stream Data.Text.Fusion 58.9 77.8317 840142256 ==_a2R5Data.Text.Fusion.Internal 11.00.0 59 0 ==_aFRUData.Text 9.70.0 52 0 text Main 8.00.0 43 227712 unsafeIndexData.Text.Array5.87.0 31 76017314 unsafeChr Data.Text.UnsafeChar 3.57.0 19 76017314 maxSizeData.Text.Fusion.Size 1.76.6 9 71476400 individualinherited COST CENTRE MODULE no.entries %time %alloc %time %alloc ticks bytes MAIN MAIN 1 0 0.00.0 100.0 100.0 0 672 mainMain 334 1 0.00.0 100.0 100.0 0 0 text Main 336 1 8.00.0 100.0 100.0 43 1821696 ==_aFRU Data.Text 34817869100 9.70.090.5 98.5 52 0 ==_a2R5 Data.Text.Fusion.Internal 35517869100 11.00.011.00.0 59 0 stream Data.Text.Fusion 34935738200 58.9 77.869.9 98.5317 6721138048 unsafeChr Data.Text.UnsafeChar 35738008657 3.57.0 3.57.0 19 608138512 unsafeIndex Data.Text.Array 35638008657 5.87.0 5.87.0 31 608138512 maxSize Data.Text.Fusion.Size 35435738200 1.76.6 1.76.6 9 571811200 lines Data.Text 346 178691 0.00.2 0.91.4 0 20013392 unsafeTail Data.Text.Unsafe 358 178690 0.00.1 0.00.1 0 5718080 iter_ Data.Text.Unsafe 359 178690 0.00.0 0.00.0 0 0 spanBy Data.Text 350 714763 0.60.4 0.91.2 3 31449600 iterData.Text.Unsafe 352 1763166 0.20.5 0.20.5 1 42315984 textP Data.Text.Internal 351 357382 0.00.2 0.20.3 0 14295280 text Data.Text.Internal 353 357381 0.20.1 0.20.1 1 11436192 null Data.Text 347 357382 0.00.0 0.00.0 0 0 decodeUtf8Data.Text.Encoding 337 1 0.00.0 0.60.0 0 0 decodeUtf8With Data.Text.Encoding 338 2 0.60.0 0.60.0 380 textP Data.Text.Internal 344 1 0.00.0 0.00.0 0 0 text Data.Text.Internal 345 1 0.0
