November 2013 - Libraries

Proposal: merge either into transformers
by Roman Cheplyaka 03 Jan '17

03 Jan '17

I propose to add the sole module of the 'either' package[1], Control.Monad.Trans.Either, to the transformers package. It provides EitherT, a very basic and fundamental data type. The difference between EitherT and ErrorT is that the latter has an Error constraint, which is used to imlement 'fail'. Note that 'either' depends on the 'semigroupoids' and 'semigroup' packages to provide appropriate instances. The proposal is not to add those instances to 'transformers' to avoid additional dependencies. The instances can then be left in the 'either' package or moved to the 'semigroupoids' and 'semigroup' packages respectively. ('semigroupoids' already depends on 'transformers', while 'semigroups' does not). Compared to the 'either' package, Show, Read, Eq and Ord instances will be dropped to keep the code Haskell2010 (those instances require FlexibleInstances, FlexibleContexts, and UndecidableInstances). The patch is attached. [*] [*] against transformers-0.3.0.0, because the darcs version is not buildable (Control/Monad/Signatures.hs is not in the repository). Roman

22 72

Proposal: Use time rather than old-time in directory
by Ian Lynagh 17 Sep '14

17 Sep '14

Hi all, This proposal is to change directory to use the time package rather than the old-time package. This will hopefully be a significant step towards actually making the transition in general. I'm not sure if POSIXTime or UTCTime is the better type to use. Does anyone know? I started off with UTCTime, then switched to POSIXTime as UTCTime didn't have a Show instance. I've since found that actually UTCTime has both Show and Read instances, but they are orphaned. So actually, if UTCTime is the better type then switching back to UTCTime would allow us to fix this comment in hpc: -- We would rather use ClockTime in Mix, but ClockTime has no Read instance and simplify the code slightly. Attached are patches for directory, hpc and ghc. Suggested discussion period: Until 5th Nov 2011. Thanks Ian

14 35

Proposal: Add Data.Semigroup to base, as a superclass of Monoid
by John Wiegley 11 Aug '14

11 Aug '14

1. I propose that we add the following package to base: http://hackage.haskell.org/packages/archive/semigroups/0.9.2/doc/html/Data-… This is somewhat in the spirit of the AMP proposal: further improving the correctness of our algebraic abstractions. 2. That we make Semigroup a superclass of Monoid, so that (minimally): class Semigroup a where (<>) :: a -> a -> a class Semigroup a => Monoid a where mempty :: a mconcat :: [a] -> a mconcat = foldr (<>) mempty mappend :: Semigroup a => a -> a -> a mappend = (<>) 3. (Optional, recommended) There are other useful functions that can be added to Semigroup, such as sconcat and times1p, but I will let Edward speak to whether those should be proposed at this time. 4. (Optional, recommended) That we fix the Monoid instance for Maybe to be: instance Semigroup a => Semigroup (Maybe a) where Just x <> Just y = Just (x <> y) _ <> _ = Nothing instance Semigroup a => Monoid (Maybe a) where mempty = Nothing For some clarification on what semigroups are and why we'd want to change Monoid, I excerpt here a selection from Brent Yorgey's "Monoids and Variations" paper: Semigroups A semigroup is like a monoid without the requirement of an identity element: it consists simply of a set with an associative binary operation.... Of course, any monoid is automatically a semigroup (by forgetting about its identity element). In the other direction, to turn a semigroup into a monoid, simply add a new distinguished element to serve as the identity, and extend the definition of the binary operation appropriately. This creates an identity element by definition, and it is not hard to see that it preserves associativity.... Adding a new distinguished element to a type is typically accomplished by wrapping it in Maybe. One might therefore expect to turn an instance of Semigroup into an instance of Monoid by wrapping it in Maybe. Sadly, Data.Monoid does not define semigroups, and has a Monoid instance for Maybe which requires a Monoid constraint on its argument type... This is somewhat odd: in essence, it ignores the identity element of [the type] and replaces it with a different one. -- John Wiegley FP Complete Haskell tools, training and consulting http://fpcomplete.com johnw on #haskell/irc.freenode.net

9 9

Re: Add laws to Alternative
by Twan van Laarhoven 29 May '14

29 May '14

On 01/11/13 12:44, Nathan van Doorn wrote: > Firstly, I don't see how IO is relevant here, it has neither a MonadPlus > instance nor an Alternative instance. You are right. I thought it was an instance with mzero=fail "foo" and mplus=catch. But I was apparently mistaken. Objection withdrawn. > Secondly, the MonadPlus laws are documented in Control.Monad to be: > > mzero >>= f = mzero > v >> mzero = mzero I missed them, because they are written in the documentation of mzero rather than the documentation of the class where I expected them. > Thirdly, the monoid laws are already documented. (<|>) must be "An associative > binary operation", and empty "The identity of <|>". These are exactly the monoid > laws. Perhaps they should be made more explicit, but that is a different issue. Missed this as well. > Fourthly, [] fulfils neither the left-distribution law or the left-catch law, > and I doubt many people would be happy to lose []'s MonadPlus instance. List does satisfy left distribution: λ> (,) <$> ([1,2] <|> [3]) <*> [4,5] [(1,4),(1,5),(2,4),(2,5),(3,4),(3,5)] λ> (,) <$> [1,2] <*> [4,5] <|> (,) <$> [3] <*> [4,5] [(1,4),(1,5),(2,4),(2,5),(3,4),(3,5)] λ> quickCheck (\x y z -> ((,) <$> (x <|> y :: [Int]) <*> (z :: [Int])) == (((,) <$> x <*> z) <|> ((,) <$> y <*> z))) +++ OK, passed 100 tests. See also http://www.haskell.org/haskellwiki/MonadPlus. Then law which it doesn't is right distribution. Consider Maybe. it does satisfies left catch but not left distribution for MonadPlus. Since mplus (Just False >>= guard) (Just True >>= guard) = Just () while mplus (Just False) (Just True) >>= guard = Nothing But for Alternative, you can't have the failure of the second argument of (<*>) depend on the first. So Maybe *does* satisfy left distribution for Alternative. IMO that makes it a good candidate law. Twan > I believe I have addressed all your issues. If I've missed something, please > point it out to me. > > Nathan. > > > On 1 November 2013 12:09, Twan van Laarhoven <twanvl(a)gmail.com > <mailto:twanvl@gmail.com>> wrote: > > On 01/11/13 11:42, Nathan van Doorn wrote: > > Proposal: add the following laws to the documentation of > Control.Applicative.__Alternative: > > * empty <*> a = empty > * f <*> empty = empty > > > These laws correspond to the laws given in MonadPlus- if you take mzero = > > empty and ap = (<*>), the ones in MonadPlus imply these- and I don't think > > this proposal should be too controversial. > > As far as I can see, the documentation for MonadPlus does not specify these > laws anywhere [1,2]. > > Consider the IO monad. These laws claim that > > launchMissiles *> fail "empty" = fail "empty" > > This is clearly *not* true. > > -- > > If we add laws, I think we should first consider the much more reasonable > monoid laws > > identity > empty <|> a = a > a <|> empty = a > associativity: > (a <|> b) <|> c = a <|> (b <|> c) > > In the MonadPlus world, the controversial part is the choice between > > left distribution > (f <|> g) <*> a = (f <*> a) <|> (g <*> a) > > or > > left catch > pure a <|> b = pure a > > Your proposal would be > > left zero > > empty <*> a = empty > right zero > > f <*> empty = empty > > And as mentioned above, right zero is problematic. The fmap version should > be okay though > > map zero > f <$> empty = empty > > > Twan > > [1] > http://hackage.haskell.org/__package/base-4.6.0.1/docs/__Control-Monad.html… > <http://hackage.haskell.org/package/base-4.6.0.1/docs/Control-Monad.html#t:M…> > [2] http://www.haskell.org/__haskellwiki/MonadPlus > <http://www.haskell.org/haskellwiki/MonadPlus> > _________________________________________________ > Libraries mailing list > Libraries(a)haskell.org <mailto:Libraries@haskell.org> > http://www.haskell.org/__mailman/listinfo/libraries > <http://www.haskell.org/mailman/listinfo/libraries> > >

9 11

Proposal: Add a strict version of <$> for monads
by Johan Tibell 22 Dec '13

22 Dec '13

Hi all, I propose we add a strict version of <$> to base: -- | A strict version of 'Data.Functor.<$>' for monads. (<$!>) :: Monad m => (a -> b) -> m a -> m b f <$!> m = do a <- m return $! f a {-# INLINE (<$!>) #-} infixl 4 <$!> It works on Monads instead of Functors as required by us inspecting the argument. This version is highly convenient if you want to work with functors/applicatives in e.g. parser and avoid spurious thunks at the same time. I realized that it was needed while fixing large space usage (but not space-leak) issues in cassava. I believe Edward Kmett discovered the need for it independently as well. Deadline: 3 weeks Cheers, Johan

13 26

Proposal: add fusion RULES for Data.Map.mapMaybe
by Akio Takano 04 Dec '13

04 Dec '13

Hi, Currently mapMaybe does not fuse at all. The attached patch implements necessary rules for fold/build fusion in both sides (the result and the second argument) to happen. When fusion does not happen, the function should behave exactly the same as before. I ran some benchmarks, testing cases where fusion does and does not happen: http://htmlpreview.github.io/?https://github.com/takano-akio/mapmaybe-bench… The benchmark code is: https://github.com/takano-akio/mapmaybe-benchmarks/blob/master/main.hs Discussion period: 2 weeks. Thank you, Takano Akio

3 3

role signatures in libraries
by Richard Eisenberg 03 Dec '13

03 Dec '13

Hi libraries, It seems that the syntax and semantics of role annotations has settled, and so it's time to add role annotations to various libraries. Here are the key ideas: tl;dr: Add nominal role annotations for abstract datatypes whose data has an invariant based on a class instance. Set and Map are top candidates. - We have three notions of equality: nominal, representational, and phantom. - Type equality as we know it is called *nominal*. - A newtype is considered *representationally* equal to its representation type. * Thus, if we have `newtype Age = MkAge Int`, `Age` and `Int` are representationally equal. - Any two types are "phantomly" equal. * See below for an illustrative example of why this notion is useful. - Every type parameter of every datatype and class is assigned a role. A parameter's role says what notion of equality must hold to show that two instantiations of the datatype are representationally equal. * Say we have `Bar a` where a's role is nominal. That means that `Bar Age` is *not* representationally equal to `Bar Int`. * Say we have `Foo a` where a's role is representational. That means that `Foo Age` is representationally equal to `Foo Int`. * Say we have `Quux a` where a's role is phantom. That means that `Quux Int` is representationally equal to `Quux Bool`. - Datatype roles are inferred to be the most permissive possible. * `data Bar a = MkBar (F a)` where `F` is a type family: a's role will be inferred to be nominal. * `data Foo a = MkFoo a`: a's role will be inferred to be representational. * `data Quux a = MkQuux String`: a's role will be inferred to be phantom. - Class roles are inferred to be nominal. * This choice follows from the idea that the dictionaries for `Ord Int` and `Ord Age` should not be assumed to be the same. - A role annotation can be used to change a parameter's role. The syntax is a top-level declaration `type role <typename> <list of roles>`. You need the extension RoleAnnotations to use these. Role names are spelled out (nominal, representational, phantom) and are in lower-case. * If we wanted, say, Quux's parameter's role to be representational, we could say `type role Quux representational`. - Role annotations must be in the same module as a type is defined in. - Roles can be left out by inserting `_`. This causes the role to be inferred. - Role annotations cannot be used to undercut type safety. They are checked to make sure they meet this requirement. - The new function `coerce` (in GHC.Exts) allows for 0-cost conversion between representationally equal types. * This means that converting [Quux Bool] to [Quux Int] is now free. This is why phantom equality is interesting. - You should consider adding a role annotation on any type that has an invariant based on class instances. * For example, the information stored in a Set is ordered according to the data's Ord instance. So, there should probably be an annotation `type role Set nominal`. * As a further example, the data stored in a Map cares about the Ord instance of its first parameter but not its second. So, we would probably want `type role Map nominal representational`. Note that the new roles stuff does *not* allow programmers to do more bad things than they could previously. A sufficiently clever use of GeneralizedNewtypeDeriving could invalidate any instance-based invariants you wished to impose. However, we now have the power to stop these leaks of abstraction. We also have, in `coerce`, an easier way to take advantage of any remaining holes, making it even more important to close the gaps. The user manual is updated with respect to all the roles stuff -- you may want to see there for more info. Please let me know if you have any questions about this! Richard

6 7

aeson and dlist in HP 2013.4.0.0
by Mark Lentczner 28 Nov '13

28 Nov '13

We seem to have several interlocking decisions to make about *aeson* and *dlist* in the platform: *Background:* *let me know if I got any of this wrong* - There is clear agreement to have *aeson* in the platform. - *aeson* 0.6.2.1 would require *dlist* & *blaze-builder* to be added. - There is a patch to *aeson*, already in head, that makes *aeson* use the builder in *bytestring*, and drop the dependency on *blaze-builder*. This is good, as a *bytestring* in GHC 7.8 will make *blaze-builder*obsolete. - *aeson* head uses a new Scientific type for arbitrary precision floating point. This will break users of the Number constructor of the Value type. However, it is believed that most users are probably using existing functions to parse to an from standard numeric types when needed, and those continue to work. Unclear how much impact this will have. Also, Scientific could be in it's own package (which would need to be added to the platform), or simply exposed from *aeson*. - *dlist* is stable, but could use some love - which was lovingly offered, and we could have a bump which adds some useful typeclass instances. - It was noted that *dlist* could be replaced by Endo, or explicit use of simple types. There didn't seem to be much support for the idea of altering *aeson* to do so. *Options for aeson:* 1. skip it in this release 2. include *aeson* 0.6.2.1 - requiring both *dlist* and *blaze-builder* 3. include *aeson* 0.6.2.x, a version with the patch that uses *bytestring*'s builder, and so require only *dlist* 4. include *aeson* 0.7.0.0 - requiring *dlist* and possibly *scientific* *Options for dlist, if required by the aeson choice, or because we now like it anyway:* 1. include *dlist* 0.5 2. include *dlist* 0.6, with new typeclass instances added *Options for scientific, if required by the aeson choice:* 1. include Scientific type in *aeson* 2. include a new *scientific* pacakge *Discussion:* I think the best option, if we are ready to embrace aeson, is jump in with both feet: aeson 0.7.0.0, & dlist 0.6. Leaving the issue of is scientific ready to be included as it's own package, or should it just be exported by aeson for now. Thoughts? — Mark

19 37

2013.4.0.0 proposal - getting to alpha
by Mark Lentczner 27 Nov '13

27 Nov '13

Here is the current status of the 2013.4.0.0 proposal. *Outstanding Issues* We have a few decisions outstanding, and I'd like to have these decided by Nov. 23 (one week). - fgl - waiting to hear if Ivan will have new version ready in time. - aeson / dlist - there are serveral options -- I'll start a separate thread on this topic. - alex / happy - I'd like to hear an "okay" from the maintainers on the major version bump *Current Proposed Versions* *GHC 7.6.3* -- same as last time (erroneously listed as a version bump last time) *Packages with no changes:* fgl, haskell-src, html, HUnit, mtl, parsec, QuickCheck, random, regex-base, regex-compat, regex-posix, split, stm, text, transformers, xhtml, zlib *Packages held back:* cgi: 3001.1.7.5 (3001.1.8.4* requires MonadCatchIO-mtl)* *Packages with minor version bumps:* HTTP: 4000.2.8 → 4000.2.9 network: 2.4.1.2 → 2.4.2.0 parallel: 3.2.0.3 → 3.2.0.4 syb: 0.4.0 → 0.4.1 unordered-containers: 0.2.3.0 → 0.2.3.3 vector: 0.10.0.1 → 0.10.9.1? primitive: 0.5.0.1 → 0.5.1.0? *Packages with major version bumps:* case-insensitive: 1.0.0.1 → 1.1.0.1 hashable: 1.1.2.5 → 1.2.1.0 GLUT: 2.4.0.0 → 2.5.0.1 GLURaw: 1.3.0.0 → 1.4.0.0 OpenGL: 2.8.0.0 → 2.9.1.0 OpenGLRaw: 1.3.0.0 → 1.4.0.0 alex: 3.0.5 → 3.1.0 *pending okay* happy: 1.18.10 → 1.19.0 *pending okay* *Cabal:* Cabal: 1.16.0 → 1.18.1.2 *consensus seems to be that this will be okay* cabal-install: 1.16.0.2 → 1.18.0.2 *Platform Packagers:* This is likely to be the last HP source release that contains all the ad hoc scripts for packaging. I canvased the group last time and as I recall, everyone said they just used the .cabal file - and no one used the scripts. I plan to replace them with a new tool built for the needs of the release, and ditch this old set of scripts. If you need 'em, speak now, or forever hold your peace! — Mark

7 11

Re: 2013.4.0.0 proposal - getting to alpha
by Ivan Lazar Miljenovic 26 Nov '13

26 Nov '13

Hurt, no; but it will make it inconsistent across platforms (as Linux distros typically just have meta-packages for the Platform, rather than an installer that can do this). On 26 November 2013 17:16, Mikhail Glushenkov <the.dead.shall.rise(a)gmail.com> wrote: > Hi, > > On Tue, Nov 26, 2013 at 6:02 AM, Ivan Lazar Miljenovic > <ivan.miljenovic(a)gmail.com> wrote: >> On 26 November 2013 15:46, Mikhail Glushenkov >> <the.dead.shall.rise(a)gmail.com> wrote: >>> [...] >>> The installer can run `ghc-pkg hide >>> Cabal-1.16` as the final step, so the user will never even see the >>> extra copy of Cabal. >> >> Is this necessary? > > Well, hiding Cabal-1.16 certainly can't hurt, can it? > > -- > () ascii ribbon campaign - against html e-mail > /\ www.asciiribbon.org - against proprietary attachments -- Ivan Lazar Miljenovic Ivan.Miljenovic(a)gmail.com http://IvanMiljenovic.wordpress.com

4 5