http://stackoverflow.com/questions/32859038/java-8-list-to-map-with-stream

IntStreams:
http://www.deadcoderising.com/2015-05-19-java-8-replace-traditional-for-loops-with-intstreams/

Characteristics:
SIZED
SUBSIZED
SORTED
ORDERED
DISTINCT
NONNULL
IMMUTABLE
CONCURRENT

characterized by Spliterator
.trySplit()

.boxed()
    ???

.sorted()
    full-barier operation (downstream ops can't start until this one ends)
.skip() and .limit()
    does not preserve the stream size :(
    .sublis() - is more effective
    .parallell() stream really utilize Spliterator and do a HUGE perf-boost
.distinct()
    .parallell().unordered() - can speed-up according to doc-n, but 10x slower than parallell and 3x to seq-l
    .parallell.ordered and .parallell.unordered a just 2 distinct algos
.concat()
    ??? temporary or terminal
    preserves SIZED if parts are SIZED
.flatMap()
    like concat but better with ORDERED streams than .concat()
    disadvantages (vs concat) are:
        - no more speedup more than 2x using parallel
        - don't preserves SIZED
        - uses more memory
        - short-circuiting (find result earlier than finishing consuming all the input)
        - .flatMap...spliterator().tryAdvance() - throws OOM because of the buffering

.orElseThrow(IllegalStateException::new)

Java9 stuff:
.ofNullable(System.getProperty("bla-bla"))
.takeWhile(predicate)
    stops once a failed-to-match element found
.dropWhile(predicate)
    -//- inverse
.iterate(3, x -> x < 100, x -> x + 3)
    java9 adds a middle/second bound-parameter