Thanks for the review, Jon. My rationale was that RuleBasedCollator is the only Collator subclass in the JDK. But thinking about it some more, I think adding an instanceof condition will be a more robust choice. So here’s a new webrev with that change:
http://cr.openjdk.java.net/~hannesw/8210683/webrev.01/ Hannes > Am 12.10.2018 um 00:36 schrieb Jonathan Gibbons <[email protected]>: > > Nice! > > FWIW, the style in langtools is to avoid the language "assert" mechanism, > because it is too difficult > to control if/when assertions might be enabled. javac provides an Assert > class, but that is not > available here, so in this case I would either remove the assert, if you > think the check is not that > important, or else use instanceof, if you think it is important. > > -- Jon > > > On 10/10/2018 02:16 AM, Hannes Wallnöfer wrote: >> Please review: >> >> Bug: https://bugs.openjdk.java.net/browse/JDK-8210683 >> Webrev: http://cr.openjdk.java.net/~hannesw/8210683/webrev.00/ >> >> Docs generated with patch applied: >> http://cr.openjdk.java.net/~hannesw/8210683/api/ >> >> Note that the new behaviour is to put the var-args signature last, whereas >> before var-args was in first position (even before the no-args signature). >> >> Thanks, >> Hannes >> >
