On 07/14/2014 01:37 PM, Paul Sandoz wrote:
On Jul 14, 2014, at 12:57 PM, Remi Forax <fo...@univ-mlv.fr> wrote:
On 07/14/2014 12:51 PM, Paul Sandoz wrote:
On Jul 12, 2014, at 5:41 PM, Remi Forax <fo...@univ-mlv.fr> wrote:
I was not able to find the answer to my question in the archive,
why Stream.concat is not implemented like this ?
@SafeVarargs
public static <T> Stream<T> concat(Stream<T>... streams) {
return Arrays.stream(streams).flatMap(Function.identity());
}
Because the capabilities and characteristics of the streams are then lost e.g.
in this case the splitting is governed by the number of streams passed in.
it seems to be a limitation of flatMap in that case, no ?
That would be much harder to optimise since each element gets mapped to a Stream of 0 or
more elements when the pipeline is executed. The operation has no "global" view
of all the streams to concatenate as all it knows is a mapping function. At the moment
flatMap is quite a simple and efficient stateless operation and i think it best it stays
that way.
e.g. imagine the case of streaming over the lines of a file and flatMapping
each line to one or more words.
Paul.
and imagine the case where you know the size of a stream returned by
flatMap, in that case,
you may want to split before pumping the values of the stream.
Rémi
