Regarding wildcards: you definitely want Predicte<? super T>, which is
what we use elsewhere. For UnaryOperator, you're stuck with invariance.
On 2/16/2016 8:32 AM, Paul Sandoz wrote:
Hi Tagir,
The implementation approach looks good. I agree with Brian on the
documentation, esp. about the connection to a for loop.
Testing-wise adding to the spliterator data sources is ok. Because you have
done that there is no need to explicitly use SpliteratorTestHelper in
IterateTest, as that is reasonably covered by SpliteratorTest. Since you have
the data providers for streams just run it through exerciseStream instead.
Regarding wild cards we could clean this up later and include other methods
[1]. It’s easy to go overboard on this, recommend sticking to ? extends/? super
where appropriate. The existing iterate methods use UnaryOperator, thus there
is no wiggle room there, am i ok with reusing that for the 3-arg method.
Regarding the existing two-args method, my inclination is to keep the implementations
separate from the 3-args, but feel free to change to using abstract spliterator (reduce
layering) using a similar code pattern (no need for "finished" or overriding
forEachRemaining). Then you can remove Streams.NONE and it’s sneaky erased usage :-)
Paul.
[1] https://bugs.openjdk.java.net/browse/JDK-8132097
On 14 Feb 2016, at 15:53, Tagir F. Valeev <amae...@gmail.com> wrote:
Hello!
I wanted to work on foldLeft, but Brian asked me to take this issue
instead. So here's webrev:
http://cr.openjdk.java.net/~tvaleev/webrev/8072727/r1/
I don't like iterator-based Stream source implementations, so I made
them AbstractSpliterator-based. I also implemented manually
forEachRemaining as, I believe, this improves the performance in
non-short-circuiting cases.
I also decided to keep two flags (started and finished) to track the
state. Currently existing implementation of infinite iterate() does
not use started flag, but instead reads one element ahead for
primitive streams. This seems wrong to me and may even lead to
unexpected exceptions (*). I could get rid of "started" flag for
Stream.iterate() using Streams.NONE, but this would make object
implementation different from primitive implementations. It would also
be possible to keep single three-state variable (byte or int,
NOT_STARTED, STARTED, FINISHED), but I doubt that this would improve
the performance or footprint. Having two flags looks more readable to
me.
Currently existing two-arg iterate methods can now be expressed as a
partial case of the new method:
public static<T> Stream<T> iterate(final T seed, final UnaryOperator<T> f) {
return iterate(seed, x -> true, f);
}
(same for primitive streams). I may do this if you think it's
reasonable.
I created new test class and added new iterate sources to existing
data providers.
Please review and sponsor!
With best regards,
Tagir Valeev.
(*) Consider the following code:
int[] data = {1,2,3,4,-1};
IntStream.iterate(0, x -> data[x])
.takeWhile(x -> x >= 0)
.forEach(System.out::println);
Currently this unexpectedly throws an AIOOBE, because
IntStream.iterate unnecessarily tries to read one element ahead.
