Set size and characteristics for Spliterator used with Tags and KeyValues

[kilink]

Tags and KeyValues were inheriting their spliterator method from Iterable, which returns a Spliterator with unknown size and no characteristics set.

[shakuzen]

Thank you for the pull request. This does seem better in principle and we should do it. I'm curious, though, if you found this due to some negative effect in your code using this.

[kilink]

Thank you for the pull request. This does seem better in principle and we should do it. I'm curious, though, if you found this due to some negative effect in your code using this.

I did not observe any actual problems due to this, but it's easy to demonstrate how it can lead to inefficiencies. It is something I stumbled across and seems to be fairly common for code that extends Iterable. I get the sense that it's almost never a good idea to rely on the default spliterator method from Iterable!

Here are a couple of examples of why it can be good to propagate the exact size:

1. Calling count on a Stream will just return the size directly if it's known, otherwise it's forced to do the reduction / iteration.
2. Calling trySplit on a Spliterator of unknown size will allocate an array of 1024 elements by default, which is likely much larger than a typical set of tags; granted, I think trySplit is more commonly used for parallel streams, but I have seen some code that makes use of trySplit even for non-parallel streams.

I didn't comb through all of the internals of Stream, but there are likely other optimizations that are made possible by propagating the exact size.

[kilink]

I forgot to mention that an additional benefit to overriding the spliterator method from Iterable, is that even though Tags has its own stream method, users may still call spliterator directly, in which case they'd also get a spliterator with no characteristics set (SORTED, DISTINCT, etc). That means such operations would also be needlessly inefficient.

[shakuzen]

Thank you for the added details. That makes sense. I was wondering if there were examples of places in code outside of this repository that Tags spliterator is being used, so I can learn about some usage in users' code that I might not have thought about. It's a bit hard to do a GitHub search for these things given the generic method names, and of course there is a lot of code that isn't in GitHub that may be using this.