Adding partitionEitherM #2641

blast-hardcheese · 2018-11-26T02:41:07Z

Used partitionEitherM in a project to ensure there weren't any odd usability concerns
Added Foldable[F].partitionEitherM[G, A, B, C]
Duplicated all partitionEither tests and modified them to test partitionEitherM
Ran sbt test
Ran sbt prePR

blast-hardcheese · 2018-11-26T03:19:26Z

Getting an unrelated error due to decimal precision in JS the JVM variants:

[info] Tests:
[info] - CommutativeGroup[BigDecimal].commutativeGroup.associative *** FAILED ***
[info]   GeneratorDrivenPropertyCheckFailedException was thrown during property evaluation.
[info]    (Discipline.scala:14)
[info]     Falsified after 96 successful property evaluations.
[info]     Location: (Discipline.scala:14)
[info]     Occurred when passed generated values (
[info]       arg0 = 5.182509090085532E-184,
[info]       arg1 = -5.036270751383713E-218,
[info]       arg2 = 2.0611576410519167E-208
[info]     )
[info]     Label of failing property:
[info]       Expected: 5.182509090085532000000002061157641E-184
[info]       Received: 5.182509090085532000000002061157640E-184

codecov-io · 2018-11-29T22:11:50Z

Codecov Report

Merging #2641 into master will decrease coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #2641      +/-   ##
==========================================
- Coverage   95.15%   95.14%   -0.02%     
==========================================
  Files         365      365              
  Lines        6774     6794      +20     
  Branches      297      314      +17     
==========================================
+ Hits         6446     6464      +18     
- Misses        328      330       +2

Impacted Files	Coverage Δ
core/src/main/scala/cats/syntax/foldable.scala	`100% <100%> (ø)`	⬆️
...patTest/src/main/scala/catsBC/MimaExceptions.scala	`0% <0%> (ø)`	⬆️
core/src/main/scala/cats/data/Kleisli.scala	`97.29% <0%> (+0.02%)`	⬆️
core/src/main/scala/cats/syntax/apply.scala	`57.14% <0%> (+32.14%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b4d1ad1...43e4276. Read the comment docs.

ceedubs · 2018-12-03T16:00:54Z

Thanks, @blast-hardcheese! This looks good to me, but introduces a binary incompatibility, so we'll have to figure out how to handle that. We'll either need to temporarily add it on as a syntax extension or wait until Cats 2.0 to merge it. cc @kailuowang @LukaJCB

LukaJCB · 2018-12-03T16:59:35Z

I think making this a syntax extension is the better path :) WDYT @blast-hardcheese?

blast-hardcheese · 2018-12-04T01:35:06Z

@LukaJCB I wouldn't be opposed to that, but that would introduce the same binary incompatibility, just for the syntax class, wouldn't it?

LukaJCB · 2018-12-04T11:51:55Z

Not quite, you can get it working by creating a new trait that gets mixed into the various imports :)
You can find a lot of examples if you search for e.g. BinCompat1 in the codebase

blast-hardcheese · 2018-12-28T20:49:15Z

Resolved conflict with another BinCompat that showed up in the interim

kailuowang

You are extending the type class instance, while normally we would just add a syntax extension to the data type. Since another binary compatible change PR #2292 is just merged, you can take advantage of the workaround structure added there and simply change your partitionEitherM accordingly and add to FoldableOps0,

blast-hardcheese · 2018-12-28T23:43:56Z

@kailuowang Understood. I was on the fence about how this method should be consumed, as I try to strike a good balance between convenience and discoverability for my teammates. Are there objections to having both?

blast-hardcheese · 2018-12-28T23:44:14Z

For the time being, I've made the changes you suggested

kailuowang · 2018-12-29T02:29:46Z

@blast-hardcheese I see, I am not against having both. A sbt fmt would probably fix the build.

ceedubs · 2019-01-12T19:02:24Z

core/src/main/scala/cats/syntax/foldable.scala

+   * res1: (List[Nothing], List[Int]) = (List(),List(4, 8, 12, 16))
+   * }}}
+   */
+  def partitionEitherM[G[_], A, B, C](fa: F[A])(f: A => G[Either[B, C]])(implicit A: Alternative[F],


This code is duplicated in two places. What if instead we added a partitionEitherM method to the Foldable companion object and had the FoldableOps0 method delegate to it? At that point, I'm not sure if it makes sense to have the FoldableOps1 extension to the type class itself; people could just call the method on the companion object if they wanted it. WDYT @kailuowang and @blast-hardcheese?

Sorry I'm sure that this is annoying. I think that Cats maintainers need to put some effort into our approach to binary compatibility and making it easier on other contributors.

Consolidating definitely felt like the right approach, but doing the same syntax lookup again while we're already working around bin compat seemed worse. I can make the change.

ceedubs · 2019-01-13T18:35:13Z

core/src/main/scala/cats/syntax/foldable.scala

+   * res1: (List[Nothing], List[Int]) = (List(),List(4, 8, 12, 16))
+   * }}}
+   */
+  def partitionEitherM[G[_], A, B, C](fa: F[A])(f: A => G[Either[B, C]])(implicit A: Alternative[F],


Would it make sense to generalize this to any Bifoldable instead of just Either?

def partitionBiffyM[G[_], H[_,_], A, B, C](fa: F[A])(f: A => G[H[B, C]])(implicit A: Alternative[F], M: Monad[G], H: Bifoldable[H]): G[(F[B], F[C])] = { import cats.instances.tuple._ implicit val mb: Monoid[F[B]] = A.algebra[B] implicit val mc: Monoid[F[C]] = A.algebra[C] F.foldMapM[G, A, (F[B], F[C])](fa)( a => M.map(f(a)){ H.bifoldMap[B, C, (F[B], F[C])](_)( b => (A.pure(b), A.empty[C]), c => (A.empty[B], A.pure(c))) } ) }

Then you could use it for the partitionEitherM use-case, but you could also use it for other Bifoldables like Tuple2:

scala> Foldable[List].partitionBiffyM(list)(a => Eval.now((a - 1, a + 1))) res5: cats.Eval[(List[Int], List[Int])] = cats.Eval$$anon$6@67bef13e scala> res5.value res6: (List[Int], List[Int]) = (List(0, 1, 2, 3),List(2, 3, 4, 5))

I'm hoping that someone else has a better name than mine 😛

This seems to suggest a broader refactor that includes partitionEither. I can generalize to partitionBiffy as well as partitionBiffyM, while providing the concretized partitionEitherM that just defers to partitionBiffyM, if that is desirable

I gave it a shot, though the examples are getting increasingly convoluted I think.

kailuowang · 2019-01-17T16:24:32Z

looks like this one can use another sbt fmt

blast-hardcheese · 2019-01-17T20:17:20Z

@kailuowang just pushed a fix -- I seem to remember somewhere seeing that it's not desired to force push to branches in cats, is this a hard rule? I'd really like to squash some of the noise, only keeping the relevant commits

kailuowang · 2019-01-17T20:31:56Z

thanks @blast-hardcheese! generally speaking, it's more preferable to not change commit history of a PR. It's easier for reviewers if your new changes come in as new commits, otherwise they will have to review everything from scratch. We simply squash everything on merge, that way the history on master is clean.

blast-hardcheese · 2019-01-17T20:34:14Z

Understood.

LukaJCB · 2019-01-17T21:23:06Z

I just want to preface this, that I really appreciate the time you take to make this PR perfect, so thanks a ton @blast-hardcheese!
I'm really sorry to continue the bikeshedding here, especially after you've already made a ton of changes to your PR, but can we name the functions partitionBifold instead?
Sounds nicer to me than bifoldable and also reflects the implementation, we partition using some structure which supports a bifold, so it seems nicer and more accurate IMO.

Again thanks for your work so far! :)

ceedubs · 2019-01-18T01:41:18Z

Agreed. Thanks a lot, @blast-hardcheese!

LukaJCB · 2019-01-18T16:10:55Z

Can you run scalafmt once more? :)

blast-hardcheese · 2019-01-18T16:20:18Z

Done

ceedubs

👍 LGTM. Thanks a lot for this really high quality PR, @blast-hardcheese!

ceedubs · 2019-01-22T03:49:19Z

core/src/main/scala/cats/syntax/foldable.scala

+   * scala> val list = List(1,2,3,4)
+   * scala> list.partitionBifold(a => (a, "value " + a.toString))
+   * res0: (List[Int], List[String]) = (List(1, 2, 3, 4),List(value 1, value 2, value 3, value 4))
+   * `Const`'s second parameter is never instantiated, so we can use an impossible type:


Should line 231 have a // at the beginning of the comment? I guess that I don't know whether or not that's necessary in a doctest example; just wanted to make sure that this code is running when tests are run.

I just assumed the parser rules were similar to literate extensions for languages. Altering line 232 to have a syntax error caused coreJVM/test:compile to fail to compile, which was good as it seems as though there are other instances of this: https://github.com/typelevel/cats/blob/774fb51/core/src/main/scala/cats/Foldable.scala#L48

I should say, I'm not opposed to changing it, just that it currently isn't causing the tool to stop compiling the other lines.

ceedubs · 2019-01-22T03:52:39Z

tests/src/test/scala/cats/tests/FoldableSuite.scala

+    }
+  }
+
+  test("Foldable#partitionEitherM consistent with List#partition") {


Ooh this is a good idea for a test 👍

LukaJCB

Awesome! Anyone have any objections to merging this as is? :)

blast-hardcheese changed the title ~~Adding partitionEither~~ Adding partitionEitherM Nov 26, 2018

blast-hardcheese force-pushed the partitionEitherM branch from a0caf63 to ec384d0 Compare November 29, 2018 22:11

ceedubs added the Binary Breaking label Dec 3, 2018

blast-hardcheese force-pushed the partitionEitherM branch from c005111 to b2d91e0 Compare December 28, 2018 20:48

blast-hardcheese force-pushed the partitionEitherM branch from b2d91e0 to 8e99343 Compare December 28, 2018 21:16

kailuowang reviewed Dec 28, 2018

View reviewed changes

blast-hardcheese force-pushed the partitionEitherM branch from 8e99343 to ec52405 Compare December 28, 2018 23:42

blast-hardcheese force-pushed the partitionEitherM branch 2 times, most recently from a32ddbe to f997256 Compare December 29, 2018 06:50

Devon Stewart added 3 commits December 28, 2018 23:12

Adding partitionEither

305b40e

Resolving 2.11 Product with Serializable inference

6c1afdc

Moving binary incompatible change out into FoldableSyntaxBinCompat1

f36ce95

blast-hardcheese force-pushed the partitionEitherM branch from f997256 to f36ce95 Compare December 29, 2018 07:14

Merge branch 'master' into partitionEitherM

959a793

kailuowang removed the Binary Breaking label Jan 12, 2019

kailuowang added this to the 1.6 milestone Jan 12, 2019

ceedubs reviewed Jan 12, 2019

View reviewed changes

Devon Stewart added 2 commits January 12, 2019 15:16

Fixing merge conflict

a23b71d

Deduplicating partitionEitherM

a522b0f

ceedubs reviewed Jan 13, 2019

View reviewed changes

Devon Stewart added 7 commits January 13, 2019 18:19

Splitting out partitionBifoldableM

f394412

Generalizing partitionBifoldable

0912a58

Typo

6dfb3b4

prePR

79b7c6f

tut fixes

3cb7be1

Accidentally shadowed fa

cc7275d

Incorrectly referenced Bitraverse

b3fb6d3

prePR

95ab386

Devon Stewart added 2 commits January 18, 2019 07:39

More doctest fixes

7f9923f

partitionBifoldable* -> partitionBifold*

bd8f0cb

LukaJCB previously approved these changes Jan 18, 2019

View reviewed changes

fmt

f4eef23

blast-hardcheese dismissed LukaJCB’s stale review via f4eef23 January 18, 2019 16:20

Doctest fixes

43e4276

ceedubs approved these changes Jan 22, 2019

View reviewed changes

LukaJCB approved these changes Jan 22, 2019

View reviewed changes

kailuowang approved these changes Jan 22, 2019

View reviewed changes

kailuowang merged commit a332935 into typelevel:master Jan 22, 2019

kailuowang added the enhancement label Jan 22, 2019

blast-hardcheese deleted the partitionEitherM branch January 22, 2019 17:55

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding partitionEitherM #2641

Adding partitionEitherM #2641

blast-hardcheese commented Nov 26, 2018 •

edited

Loading

blast-hardcheese commented Nov 26, 2018 •

edited

Loading

codecov-io commented Nov 29, 2018 •

edited

Loading

ceedubs commented Dec 3, 2018

LukaJCB commented Dec 3, 2018

blast-hardcheese commented Dec 4, 2018

LukaJCB commented Dec 4, 2018

blast-hardcheese commented Dec 28, 2018

kailuowang left a comment

blast-hardcheese commented Dec 28, 2018

blast-hardcheese commented Dec 28, 2018

kailuowang commented Dec 29, 2018 •

edited

Loading

ceedubs Jan 12, 2019

blast-hardcheese Jan 12, 2019

ceedubs Jan 13, 2019

blast-hardcheese Jan 13, 2019

blast-hardcheese Jan 14, 2019

kailuowang commented Jan 17, 2019

blast-hardcheese commented Jan 17, 2019

kailuowang commented Jan 17, 2019 •

edited

Loading

blast-hardcheese commented Jan 17, 2019

LukaJCB commented Jan 17, 2019

ceedubs commented Jan 18, 2019

LukaJCB commented Jan 18, 2019

blast-hardcheese commented Jan 18, 2019

ceedubs left a comment

ceedubs Jan 22, 2019

blast-hardcheese Jan 22, 2019

blast-hardcheese Jan 22, 2019

ceedubs Jan 22, 2019

LukaJCB left a comment

Adding partitionEitherM #2641

Adding partitionEitherM #2641

Conversation

blast-hardcheese commented Nov 26, 2018 • edited Loading

blast-hardcheese commented Nov 26, 2018 • edited Loading

codecov-io commented Nov 29, 2018 • edited Loading

Codecov Report

ceedubs commented Dec 3, 2018

LukaJCB commented Dec 3, 2018

blast-hardcheese commented Dec 4, 2018

LukaJCB commented Dec 4, 2018

blast-hardcheese commented Dec 28, 2018

kailuowang left a comment

Choose a reason for hiding this comment

blast-hardcheese commented Dec 28, 2018

blast-hardcheese commented Dec 28, 2018

kailuowang commented Dec 29, 2018 • 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

kailuowang commented Jan 17, 2019

blast-hardcheese commented Jan 17, 2019

kailuowang commented Jan 17, 2019 • edited Loading

blast-hardcheese commented Jan 17, 2019

LukaJCB commented Jan 17, 2019

ceedubs commented Jan 18, 2019

LukaJCB commented Jan 18, 2019

blast-hardcheese commented Jan 18, 2019

ceedubs left a 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

LukaJCB left a comment

Choose a reason for hiding this comment

blast-hardcheese commented Nov 26, 2018 •

edited

Loading

blast-hardcheese commented Nov 26, 2018 •

edited

Loading

codecov-io commented Nov 29, 2018 •

edited

Loading

kailuowang commented Dec 29, 2018 •

edited

Loading

kailuowang commented Jan 17, 2019 •

edited

Loading