Add Alt, Plus and Alternative specs #197

safareli · 2016-10-27T13:59:58Z

Specs added:

Alt
Plus
Alternative
~~MonadZero~~
~~MonadOr~~
~~MonadPlus~~

TODO:

add laws in laws/*
update test
~~add mPlus and mOrElse ?~~

fixes #187

Monad{Zero,Or,Plus} are removed from this PR (reason)

safareli · 2016-10-27T14:01:49Z

README.md

+the [Applicative](#applicative) and [Plus](#plus) specifications.
+
+1. `x.ap(f.alt(g))` is equivalent to `x.ap(f).alt(x.ap(g))` (distributivity)
+1. `M.pempty().ap(f)` is equivalent to `M.pempty()` (annihilation)


In purescript it's in different order and i think it's correct, so I have opened issue purescript-control#34

Because of Applicative's interchange law they are equivalent so i have sticked to purescript's version

safareli · 2016-10-27T14:11:29Z

Also we can use zero instead of pempty (from semigroupoids.Plus)

If you like zero better than pempty 👍
if you don't like zero and prefer pempty 👎

make it consistant with purescript version, even tho they are equivalent because of Applicative's interchange law

garyb · 2016-10-28T18:48:23Z

👍 for zero - we don't use that to avoid the mempty situation in PureScript only because it's used for yet another "zero" - that of Semiring 😄

safareli · 2016-10-28T19:18:39Z

Will change then! and if in future we add Semiring to FL, as it is subclass of Monoid than we would just use monoids empty for it.

garyb · 2016-10-28T19:23:11Z

Well, this is a discussion for another day, but it's not really a subclass of Monoid as there are two possible Monoids for it: (+) and 0, and (*) and 1. We have Additive and Multiplicative newtypes for those Monoids for any Semiring... not sure what the JS answer to that would be!

- move patches and test structures into `internal` - `Compose` and patching of `Array` form `laws/traversable` - `equality`, `Sum` and patching of `String` form `id_test` - `Id` from `id` - rename `id_test` to `test` as it's not just testing `Id` but `Maybe` and `Sum`

rpominov · 2016-10-28T20:31:54Z

The PR LGTM overall 👍

SimonRichardson · 2016-10-28T20:52:21Z

internal/maybe.js

+    Just: _ => this,
+    Nothing: () => b,
+  });
+};


I wonder if we just import fantasy-options, I know people don't want more dependencies, but I don't want to envision lots of Maybes out in the wild. Everyone different to each other...

This is something that I have struggled to understand myself. There doesn't need to be multiple option/maybe/either implementations; and, I don't know why we can't specify what option/maybe/either is once and for all. Half of the ones out in the wild are broken anyway. Mostly they're not functors, but they claim to be. That's a worse situation than anything else. But the data types are simple and don't necessitate reinventing the wheel in every sub-community.

Not being able to speak about option/maybe/either also complicates the spec in ways that I think are more painful than forcing one data type onto people. Look at ChainRec. As far as I understand types, you can't really implement it. In order for it to be implementable, you need higher rank polymorphism. HIgher rank polymorphism is something that very few languages/tools support–flow and typescript do not support it. I don't think we gain much as a spec by forcing higher rank polymorphism on people. It's been hard enough to communicate higher kinded polymorphism.

If we can just speak about option/maybe/either: we won't have an unimplementable algebra, ChainRec will be easier to understand, and we can bring in more useful algebras–Filterable, Witherable, Choice, CoChoice, Decidable to name a few.

Here is my responce to this issue (scroll down)

I'm sorry. I was venting. This PR is not the appropriate place for it. Do not let my comment stop progress of this PR. If it's something we are going to address, we should do it in another issue.

I feel the same way though :-(

I think we can discuss this in #200

SimonRichardson · 2016-10-28T20:54:14Z

laws/alt.js

+  a[map](f)[alt](b[map](f))
+);
+
+module.exports = {associativity, distributivity};


SimonRichardson · 2016-10-28T20:54:23Z

laws/alternative.js

+  T[zero]()
+);
+
+module.exports = {distributivity, annihilation};


SimonRichardson · 2016-10-28T20:54:32Z

laws/monador.js

+  T[of](a)
+);
+
+module.exports = {leftCatch};


SimonRichardson · 2016-10-28T20:55:09Z

laws/monadplus.js

+  x[chain](f)[alt](y[chain](f))
+);
+
+module.exports = {distributivity};


SimonRichardson · 2016-10-28T20:55:16Z

laws/monadzero.js

+  T[zero]()
+);
+
+module.exports = {annihilation};


SimonRichardson · 2016-10-28T20:55:22Z

laws/plus.js

+  T[zero]()
+);
+
+module.exports = {rightIdentity, leftIdentity, annihilation};


SimonRichardson · 2016-10-28T20:56:32Z

Just one comment otherwise 🎉

safareli · 2016-10-28T21:10:22Z

That's interesting point, suppose we depend on fantasy-id, fantasy-options, fantasy-compose ...

What would happen if someone adds other spec? how do you see for example process of adding some spec Foo with method foo?

Also fantasy-* libs have test on there own, we would test it twice? and they also depend on FL.

safareli · 2016-10-28T21:15:22Z

Possibly solution is to completely remove test from FL, but in this case fantasy-id, fantasy-maybe ... should be always up to date and if we have a PR to add some spec to FL, before we land it, PRs for fantasy-* should be opened, to ensure PR in FL is correct, then PR in FL could be merged and version bumped. After which PRs in fantasy-* could update FL and merge as well.

safareli · 2016-10-29T15:54:11Z

one issue is that Maybe is not MonadPlus but is MonadOr

mplus a b >>= f = mplus (a >>= f) (b >>= f)
a = Just 0
b = Just 1
f 0 = Nothing
f 1 = Just 1

so this will fail:

test(x => monadPlus.distributivity(equality)(Maybe.Just(0))(Maybe.Just(1))(
  a => a === 0 ? Maybe.Nothing : Maybe.Just(a)
))

Array is MonadPlus, but not MonadOr, so I have updated Array patch to test it against monadPlus laws.

- fix Maybe equals - make Array Alternative and Monad - test Array for MonadPlus - test Maybe for MonadOr

safareli · 2016-10-31T10:16:39Z

At this point it's ready to be merged, But there is one thing, which I don't quite like:

If some structure has of we know that it's Apply and if it also has chain than it must be a Monad there are no other choices. on the other hand if structure has of , chain and alt than it must be MonadZero. Problem is that it could also be a MonadPlus or MonadOr. so we have 3 specs Monad{Zero,Plus,Or} and they are not distinguishable from each other.

I think it could be problematic in js (because of lack of type system), but it could also be problematic even in purescript?

f :: (Monad___ m) => m a ->  m a -> (a-> m b) -> m b
f m1 m2 g= (m1 `alt` b2) >>= g

I don't have strong opinion on it, maybe it's not a big dial and it's actually ok, but wanted to point it out. Also I would love to hear what @garyb and @paf31 think on it as in PS there is not MonadOr (yet).

-- update

in MonadPlus reform proposal MonadPlus is defining mplus and MonadOr is defining morelse, it's good beacause user of those methods knows how would it behave exactly, and maybe it's the way to go?

morelse and mplus would be aliases of alt if structure supports only one of them, but some could support both MonadPlus and MonadOr:

instance MonadPlus Maybe where
   mplus (Just a) Nothing = Just a
   mplus Nothing (Just a) = Just a
   mplus _ _ = Nothing

instance MonadOr Maybe where
  -- this is same as definition in Alternative
   morelse (Just a) _ = Just a
   morelse _ b = b

as well as []

instance MonadOr [] where
   morelse [] b = b
   morelse a b = a
instance MonadPlus [] where
  -- this is same as definition in Alternative
   mplus  = (++)

from MonadPlus reform proposal#Instances of both

joneshf · 2016-10-31T14:17:43Z

If some structure has of we know that it's Apply and if it also has chain than it must be a Monad there are no other choices. on the other hand if structure has of , chain and alt than it must be MonadZero

Is that true? It seems like you can have something tha is Applicative and Chain, but doesn't satisfy the Monad laws. Similarly for MonadZero.

image was squashed because of `px` sizing.

safareli · 2016-10-31T15:31:58Z

@joneshf technically one could make Validation a Chain but it's sort of cheating.

So to be precise if structure is Applicative(of,ap) and Chain(chain) than we have two choices

Applicative+Chain
Applicative+Chain+Monad

if structure is Applicative(of,ap) Alternative(alt,zero) and Chain(chain) then we have 4 choices:

Applicative+Alternative+Chain
Applicative+Alternative+Chain+MonadZero
Applicative+Alternative+Chain+MonadZero+MonadOr
Applicative+Alternative+Chain+MonadZero+MonadPlus

rpominov · 2016-10-31T15:58:13Z

I wouldn't care too much about detection of supported algebras by presence of methods. If this will become a real problem we can solve it for example by requiring values to have fantasy-land/algebras property that points to an array like ['Funcor', 'Apply', 'Applicative'].

But we should think about structures that may support both MonadPlus and MonadOr, as far as I understand.

safareli · 2016-10-31T16:48:58Z

But we should think about structures that may support both MonadPlus and MonadOr, as far as I understand.

Exactly.

In that case we should have separate methods for MonadPlus and MonadOr. in the haskell proposal names: mplus and morelse are used, if we use them I think they should be camelCase: mPlus, mOrElse.

if structure is MonadPlus or MonadOr it could define alt using mPlus or mOrElse.

T.prototype[fl.mOrElse] = T.prototype[fl.alt] = function(a) { .... }
T.prototype[fl.mPlus] = T.prototype[fl.alt] = function(a) { .... }

do others agree on the names?

safareli · 2016-11-01T15:49:08Z

I think it's ready for final review and then it could be merged 🎉 🌮

davidchambers · 2016-11-01T23:09:40Z

README.md

@@ -25,7 +28,7 @@ structures:
 * [Bifunctor](#bifunctor)
 * [Profunctor](#profunctor)

-<img src="figures/dependencies.png" width="863" height="347" />
+<img src="figures/dependencies.png" width="100%" />


I think this will reintroduce the problem @rpominov fixed in #119.

Will fix that. We definitly need CONTRIBUTING.md

on github the width of README is fixed but for example on npm it looks bad because max width of readme area is 727px and if you are on small screan than it's worth.

Yeah, same issue with mobile version of GitHub for example. I wish we could find a fix without this side effect, but in the current situation having links to point to the right locations is more important imo.

Ok will change it as it was before.

But we could move the image to the end of README and have a link in it's place where it was before (next to list of specs) and also a link back.

+1 for moving to the end.

you can check rendered image here

rpominov · 2016-11-01T23:27:15Z

README.md

+#### `zero` method
+
+```hs
+zero :: Plus x => () -> x


Should it be Plus x => () -> x a? Also maybe use p instead of x?

In that case we should also add forall a I think, but i'm not familiar with that. maybe @joneshf could help here?

I think forall is kinda implicit in our loosely specified type system. What I was trying to say is that x is not a type here, it's a type constructor, so it cannot be in that position in signature. x a is a type on the other hand.

I would use f a. f to be consistent with other types

rpominov

LGTM

safareli · 2016-11-02T18:11:19Z

@SimonRichardson @davidchambers I think it's ready to be merged ⏲

SimonRichardson · 2016-11-02T19:05:56Z

I'll check tomorrow if that's ok

On Wed, 2 Nov 2016, 18:11 Irakli Safareli, notifications@github.com wrote:

@SimonRichardson https://github.com/SimonRichardson @davidchambers
https://github.com/davidchambers I think it's ready to be merged ⏲

—
You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub
#197 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ACcaGFxlV8irMSnODLUh-iPfwfuL6L4tks5q6NJKgaJpZM4KiYeK
.

safareli · 2016-11-04T09:02:37Z

⏰ a friendly reminder

SimonRichardson · 2016-11-04T12:28:24Z

@safareli sorry, I looked at this yesterday and forgot to 👍

MERGE IT
🌮

davidchambers · 2016-11-04T13:38:15Z

Would you mind squashing the commits, @safareli? It would be nice to avoid mentioning MonadZero, MonadOr, and MonadPlus in the commit history to avoid confusion.

safareli · 2016-11-04T13:57:09Z

I think this commits might be useful in future for reference. And as I remember before my committee where Squash and Merged. can we do it again? if yes then commit message will be

Add Alt, Plus and Alternative specs (#197)

And for it's description we could just say that tests where refactored a bit:

- refactor tests
- 'id.js' moved to 'internal/'

if you want to just Rebase and Merge I could squash changes but the monad zero... stuff will be lost.

davidchambers · 2016-11-04T14:04:19Z

Thanks for updating the pull request's title and providing an appropriate commit message. :)

Shall we release v2.1.0?

safareli · 2016-11-04T14:41:12Z

I think yes 🎈

rpominov · 2016-11-05T15:01:11Z

Yea, releasing a new version would be nice. I need it to add this algebras to static-land.

* 'master' of github.com:fantasyland/fantasy-land: (29 commits) Version 2.1.0 Add Alt, Plus and Alternative specs (fantasyland#197) Use uppercase letters for Type representatives in laws (fantasyland#196) Fix id_test and argument order in laws (fantasyland#193) Version 2.0.0 Another go at updating dependencies (fantasyland#192) release: integrate xyz (fantasyland#191) test: remove unnecessary lambdas (fantasyland#190) require static methods to be defined on type representatives (fantasyland#180) lint: integrate ESLint (fantasyland#189) Enforce parametricity (fantasyland#184) readme: tweak signatures to indicate that methods are not curried (fantasyland#183) Fix reduce signature to not use currying (fantasyland#182) Link to dependent specifications (fantasyland#178) Add Fluture to the list of implementations (fantasyland#175) laws/functor: fix composition (fantasyland#173) laws/monad: fix leftIdentity (fantasyland#171) Minor version bump bower: add bower.json (fantasyland#159) Fix chainRec signature to not use currying (fantasyland#167) ...

safareli commented Oct 27, 2016

View reviewed changes

safareli force-pushed the alternative branch from 3da8723 to 876d09e Compare October 28, 2016 17:45

safareli mentioned this pull request Oct 28, 2016

Add spec for Alt, Plus, Alternative #187

Closed

safareli added 2 commits October 28, 2016 22:15

add Alternative typeclasses in readme and graph

27fe5ca

reorder 'ap' arguments in Alternative's annihilation laws

0903767

make it consistant with purescript version, even tho they are equivalent because of Applicative's interchange law

safareli added 2 commits October 28, 2016 23:19

add laws and update index.js

68e6b42

rename pempty to zero

69655f7

safareli force-pushed the alternative branch from 876d09e to 58212d7 Compare October 28, 2016 19:41

safareli force-pushed the alternative branch from 58212d7 to 98e3634 Compare October 28, 2016 19:56

SimonRichardson reviewed Oct 28, 2016

View reviewed changes

SimonRichardson approved these changes Oct 28, 2016

View reviewed changes

fix MonadOr and MonadPlus implementations

2863e8c

- fix Maybe equals - make Array Alternative and Monad - test Array for MonadPlus - test Maybe for MonadOr

safareli mentioned this pull request Oct 30, 2016

move tests to corresponding structures? #200

Open

safareli mentioned this pull request Oct 31, 2016

Draw a MonadZero out of MonadPlus? purescript/purescript-control#24

Closed

fix squashed graph image

05f01ac

image was squashed because of `px` sizing.

safareli force-pushed the alternative branch from 243ab92 to 00fabbb Compare November 1, 2016 15:11

remove monad{Zero,Plus,Or} specs

cb5977d

safareli force-pushed the alternative branch from 00fabbb to cb5977d Compare November 1, 2016 15:12

use A instead of M in M.zero()

7e661ed

davidchambers reviewed Nov 1, 2016

View reviewed changes

rpominov reviewed Nov 1, 2016

View reviewed changes

update signature of zero

09798ee

rpominov approved these changes Nov 2, 2016

View reviewed changes

use fixed size for dependencies image

05fe7c4

safareli force-pushed the alternative branch from 11b4ada to 05fe7c4 Compare November 2, 2016 14:35

prefix 'id.js' with 'internal/' in readme

5edc75c

safareli changed the title ~~Add Alternative specs~~ Add Alt, Plus and Alternative specs Nov 4, 2016

davidchambers merged commit 9ca844c into fantasyland:master Nov 4, 2016

safareli deleted the alternative branch November 4, 2016 14:37

rpominov mentioned this pull request Nov 5, 2016

Alt and friends fantasyland/static-land#33

Merged

davidchambers mentioned this pull request Nov 5, 2016

fantasy-land@2.1.0 sanctuary-js/sanctuary-type-classes#12

Merged

This was referenced Nov 8, 2016

Expose non monadic version of Task origamitower/folktale#52

Open

Expose non monadic version of Future fluture-js/Fluture#44

Closed

safareli mentioned this pull request Nov 21, 2016

Consider integrating type class morphisms briancavalier/creed#6

Open

Add Alt, Plus and Alternative specs #197

Add Alt, Plus and Alternative specs #197

Conversation

safareli commented Oct 27, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

safareli commented Oct 27, 2016 • edited Loading

garyb commented Oct 28, 2016

safareli commented Oct 28, 2016 • edited Loading

garyb commented Oct 28, 2016

rpominov commented Oct 28, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

SimonRichardson commented Oct 28, 2016

safareli commented Oct 28, 2016

safareli commented Oct 28, 2016 • edited Loading

safareli commented Oct 29, 2016 • edited Loading

safareli commented Oct 31, 2016 • edited Loading

joneshf commented Oct 31, 2016

safareli commented Oct 31, 2016

rpominov commented Oct 31, 2016 • edited Loading

safareli commented Oct 31, 2016

safareli commented Nov 1, 2016

I think it's ready for final review and then it could be merged 🎉 🌮

Choose a reason for hiding this comment

safareli Nov 2, 2016 • edited Loading

Choose a reason for hiding this comment

safareli Nov 2, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rpominov left a comment

Choose a reason for hiding this comment

safareli commented Nov 2, 2016

SimonRichardson commented Nov 2, 2016

safareli commented Nov 4, 2016

SimonRichardson commented Nov 4, 2016

davidchambers commented Nov 4, 2016

safareli commented Nov 4, 2016

davidchambers commented Nov 4, 2016

safareli commented Nov 4, 2016

rpominov commented Nov 5, 2016

safareli commented Oct 27, 2016 •

edited

Loading

safareli commented Oct 27, 2016 •

edited

Loading

safareli commented Oct 28, 2016 •

edited

Loading

safareli commented Oct 28, 2016 •

edited

Loading

safareli commented Oct 29, 2016 •

edited

Loading

safareli commented Oct 31, 2016 •

edited

Loading

rpominov commented Oct 31, 2016 •

edited

Loading

safareli Nov 2, 2016 •

edited

Loading

safareli Nov 2, 2016 •

edited

Loading