RFR 8071479: Stream and lamdification improvements to j.u.regex.Matcher

[Paul Sandoz]

On Feb 12, 2015, at 9:43 AM, Peter Levart <peter.levart at gmail.com> wrote:

> 
> On 02/11/2015 08:23 PM, Stuart Marks wrote:
>>> 1.1) Change the specification of Matcher.results to reset the stream before matching, making it consistent with the replace* methods. 
>> 
>> I'm not sure about this. The current replaceAll/replaceFirst methods reset the matcher before doing any matching, so the lambda-based overloads should do the same. 
>> 
>> However, the model for 
>> 
>>     Stream<MatchResult> results() 
>> 
>> seems to me to be a stream of matches that would be returned by successive calls to find(). (Indeed, that's how it's implemented.) The no-arg find() call doesn't reset the Matcher, and it respects the existing region of the Matcher. I think results() should do the same.
> 
> Hi,
> 
> What about two methods?
> 
> Stream<MatchResult> remainingResults(); // doesn't reset the Matcher
> Stream<MatchResult> [all]results(); // resets Matcher and calls remainingResults()
> 

I would prefer to stick with just one, given that it is very easy to reset the matcher, and the most common case is to start with a new matcher.

Paul.