Spliterator issue?

[Paul Sandoz]

Hi Ben,

I think we can go ahead and commit this but i would like to iterate further on it. Did you see my email on testing? I think we can improve the scope of testing by copying the example i presented and we can widen the test data, for example we should have 0 length strings and regexes that match no input, just to test empty data.

I can commit tomorrow if you like. I may tidy up the javadoc a little and add some extra characteristics.

Paul.

On Mar 24, 2013, at 1:11 PM, Ben Evans <benjamin.john.evans at gmail.com> wrote:

> OK, looks like a Heisenbug.
> 
> Cleaning up & rebuilding from scratch seems to have fixed it. Thanks
> to Boaz for doing a quick test for me.
> 
> New webrev at: http://www.java7developer.com/webrev-kittylyst.003/
> 
> Thanks,
> 
> Ben
> 
> On Sat, Mar 23, 2013 at 3:25 PM, Ben Evans
> <benjamin.john.evans at gmail.com> wrote:
>> Hi,
>> 
>> I've rewritten the MatcherIterator for the Pattern point
>> lambdafication to properly cache values and have an idempotent
>> hasNext(), like this:
>> 
>>    private static class MatcherIterator implements Iterator<CharSequence> {
>>        private final Matcher curMatcher;
>>        private final CharSequence input;
>>        private int current = 0;
>>        private CharSequence nextElement = null;
>>        private boolean valueReady = false;
>> 
>>        MatcherIterator(CharSequence in, Matcher m) {
>>            input = in;
>>            curMatcher = m;
>>        }
>> 
>>        public void accept(CharSequence t) {
>>            valueReady = true;
>>            nextElement = t;
>>        }
>> 
>>        public CharSequence next() {
>>            if (!valueReady && !hasNext())
>>                throw new NoSuchElementException();
>>            else {
>>                valueReady = false;
>>                return nextElement;
>>            }
>>        }
>> 
>>        public boolean hasNext() {
>>            if (!valueReady) {
>>                if (current == input.length()) return false;
>>                if (curMatcher.find()) {
>>                    nextElement = input.subSequence(current, curMatcher.start());
>>                    current = curMatcher.end();
>>                    valueReady = true;
>>                } else {
>>                    nextElement = input.subSequence(current, input.length());
>>                    current = input.length();
>>                    valueReady = true;
>>                }
>>            }
>>            return true;
>>        }
>>    }
>> 
>>    public Stream<CharSequence> splitAsStream(final CharSequence input) {
>>        return Streams.stream(Spliterators.spliteratorUnknownSize(new
>> MatcherIterator(input, p.matcher(input)), Spliterator.ORDERED));
>>    }
>> 
>> which seems to behave the way that I expect, except that calling
>> splitAsStream() causes a NoSuchElementException to always be thrown.
>> 
>> Has something changed in the implementation of Spliterator which
>> causes this to happen?
>> 
>> Thanks,
>> 
>> Ben