updated to 2.13-M5

[kailuowang]

Hopefully we can merge this one quickly and publish 1.5

[codecov-io]

Codecov Report

Merging #2589 into master will increase coverage by 0.26%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #2589      +/-   ##
==========================================
+ Coverage    94.9%   95.17%   +0.26%     
==========================================
  Files         361      361              
  Lines        6615     6653      +38     
  Branches      277      293      +16     
==========================================
+ Hits         6278     6332      +54     
+ Misses        337      321      -16

Impacted Files	Coverage Δ
laws/src/main/scala/cats/laws/MonadLaws.scala	100% <0%> (ø)	⬆️
core/src/main/scala/cats/data/Ior.scala	98.54% <0%> (+0.01%)	⬆️
core/src/main/scala/cats/data/Chain.scala	99.58% <0%> (+0.01%)	⬆️
core/src/main/scala/cats/Eval.scala	98.82% <0%> (+0.04%)	⬆️
core/src/main/scala/cats/data/IorT.scala	97.79% <0%> (+0.04%)	⬆️
core/src/main/scala/cats/data/NonEmptyChain.scala	99.07% <0%> (+0.07%)	⬆️
...ore/src/main/scala/cats/data/NonEmptyMapImpl.scala	96.15% <0%> (+0.15%)	⬆️
.../src/main/scala/cats/data/RepresentableStore.scala	87.5% <0%> (+4.16%)	⬆️
core/src/main/scala/cats/syntax/either.scala	99.26% <0%> (+12.5%)	⬆️
testkit/src/main/scala/cats/tests/CatsSuite.scala	70% <0%> (+36.66%)	⬆️

---

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 a26869d...4b8db50. Read the comment docs.

[erikerlandson]

+1, this is on the path for algebra and spire

[ChristopherDavenport]

Looks like some sort of scalatest or scalactic bug.

[SethTisue]

at a glance, all that spew in the log looks like scalatest/scalatest#1423, but I think the actual failure is something else?

[kailuowang]

The previous error was fixed, now I have a randomly occurring error (roughly 50% of the time) on scalajs

[info] - Try with Throwable.monadError.applicativeError recover consistent with recoverWith *** FAILED ***
[info] UndefinedBehaviorError was thrown during property evaluation.
[info] Message: An undefined behavior was detected: scala.runtime.Statics$PFMarker$@26 is not an instance of scala.util.Try
[info] Occurred when passed generated values (
[info] arg0 = Failure(java.lang.Error),
[info] arg1 =
[info] )

[erikerlandson]

log spew might be a problem unto itself here?
The job exceeded the maximum log length, and has been terminated.

[kailuowang]

Minimized scalajs error example here
I am going to mark this as blocked by scala-js/scala-js#3480 and scalatest/scalatest#1423

[kailuowang]

turns out to be a scala.util.Failure regression, reported by @sjrd scala/bug#11242

[SethTisue]

can it be worked around for now?

[kailuowang]

@SethTisue the scala.util.Failure regression can probably be circumvented by disabling the test on 2.13 temporarily. I am not sure about the scalatest output one. I'll poke around scalatest settings.

[erikerlandson]

just one test failing for (2.13, js_build false) in OpSuite

[info] OpSuite:
[info] - Op[Function1, Char, Int].eq.antisymmetry
[info] - Op[Function1, Char, Int].eq.reflexitivity
[info] - Op[Function1, Char, Int].eq.symmetry
[info] - Op[Function1, Char, Int].eq.transitivity
[info] - Eq[Op[Function1, Char, Int]].serializable.can serialize and deserialize *** FAILED ***
[info]   ClassCastException was thrown during property evaluation.
[info]     Message: cannot assign instance of java.lang.invoke.SerializedLambda to field org.scalacheck.Gen$$anon$5.f$5 of type scala.Function2 in instance of org.scalacheck.Gen$$anon$5
[info]     Occurred when passed generated values (
[info]   
[info]     )
[info] - Op[Function1, Char, Int].category.category left identity
[info] - Op[Function1, Char, Int].category.category right identity
[info] - Op[Function1, Char, Int].category.compose associativity
[info] - Category[Op[Function1, ?, ?]].serializable.can serialize and deserialize

[SethTisue]

looks like typelevel/scalacheck#427 (comment) / scala/scala-dev#562

[kailuowang]

the fix to this one is to turn fork in test back to true, it was turned to false as a workaround for the scalatest issue. So, if you allow me,
(╯°□°）╯︵ ┻━┻

[erikerlandson]

can you just disable this single test using BuildInfo.scalaVersion != "2.13.0-M5" like you did with the other test in TrySuite?

[kailuowang]

I'd like to turn fork in test back to true, since that's better memory wise. Cheeseng is looking at scalatest/scalatest#1423 so I am hopeful.

[SethTisue]

I'd like to turn fork in test back to true, since that's better memory wise

I'm mostly just curious, but how come it isn't better to just crank up -Xmx some more? I don't recall seeing other projects setting fork to true for this reason (unless they were doing weird things with threads and classloaders, something like that, which I wouldn't expect in this repo)

[kailuowang]

@SethTisue good question! Due to the need to generate coverage report, we need to run test from root module and congregate coverage reports (although that may be hackable). Since each test run classloader is not garbage collected, it's probably better memory wise that each module's test is run in a fresh forked jvm.

[erikerlandson]

🎉

[ceedubs]

Nice. Thank you, @kailuowang and everyone else who helped make this happen!

[ceedubs]

We might want to open an issue in this repo to remind ourselves to remove this hack.

[erikerlandson]

done: #2601

[kailuowang]

thanks @erikerlandson