Infinite loop in TableStepperBase on trySplit

[lrytz]

This hangs

scala> collection.mutable.HashSet("1", "2").stepper.trySplit

here: https://github.com/scala/scala/blob/63769aff4d7301b595d5b3cd71a28c31313ff408/src/library/scala/collection/convert/impl/TableStepper.scala#L58

[lrytz]

cc @Ichoran

[lrytz]

(found while porting tests over from java8-compat, https://github.com/scala/scala/compare/2.13.x...lrytz:java8-tests?expand=1)

[SethTisue]

@lrytz is 2.13.0 a reasonable milestone for this?

[Ichoran]

I'll test the trySplits tonight. I think fixing them all will be practical in a short period of time.

[Ichoran]

Verified, and I think there are correctness issues with all the other set/map splitters. The others all seem fine. I'll investigate shortly.

[Ichoran]

BitSets also keep returning empty steppers instead of eventually giving null on splits. Bad practice, though the Java API tolerated it last I checked.

[lrytz]

I think we don't need to delay RC1 over these bugs, but should fix them for RC2 / .0

[Ichoran]

Yeah, okay. I've fixed this one; will submit one PR to fix both the TableStepper bug and the ChampStepper one (it skips elements).

[Ichoran]

Have a fix in https://github.com/Ichoran/scala/tree/issue-11468

Is 2.13.x still what I should rebase against?

[SethTisue]

yes

[lrytz]

@Ichoran I included your commit in my PR with the new tests: scala/scala#7970. It fixes the issues I observed. If you have time to review this one, that would be great :)