Fix bug in MapNode

[aljoscha]

MapNode is responsible for PlainMapOperatorBase and not for
MapOperatorBase, but this is a mess right now.

Maybe we should think about cleaning the situation a bit. CollectorMap is the same as FlatMap, only the function in the interface is called map instead of flatMap. I know that this is because of historic reasons but it is still a mess because we handle all three types through the layers. For example, some code in PactCompiler looks like this:

else if (c instanceof MapOperatorBase) {
    n = new CollectorMapNode((MapOperatorBase<?>) c);
}
else if (c instanceof PlainMapOperatorBase) {
    n = new MapNode((PlainMapOperatorBase<?>) c);
}
else if (c instanceof FlatMapOperatorBase) {
    n = new FlatMapNode((FlatMapOperatorBase<?>) c);
}

[fhueske]

I had a fix for this in #542 as well ;-)

[fhueske]

Correct me if I am wrong, but won't this break with the Record API where Map extends MapOperatorBase?

[aljoscha]

Yes, that's what I meant by historic reasons. We could change FlatMap and call the method there map() as well...

[fhueske]

My fix removed the method as it was done before the merge and the build passed Travis.

[aljoscha]

This works as well, because MapOperatorBase is converted to a CollectorMapNode, not this MapNode here.

[fhueske]

Alright then ;-)

[StephanEwen]

I agree that this needs urgent cleanout. I'll try and look over the PRs
soon...

[rmetzger]

Damn.
I suggest to have faster merge cycles for JAPI related (notice the new label) pull requests. As there are more and more developers working on it, we should sort out bugs as fast as possible.

Can we agree to a
2 x +1 (from a committer or @markus-h) AND no complaints ==> merge
voting scheme ?

[fhueske]

I agree if the two +1 do not include the PR author.

[rmetzger]

Okay. Makes sense.

[fhueske]

I agree with cleaning up the situation. It also took me quite some time to figure out how the different MapOperators relate to each other...

[fhueske]

The bug was fixed by another commit.
I'll open a new issue for the clean-up discussion and close this PR.