hg: jdk8/tl/jdk: 7126277: Alternative String hashing implementation

Doug Lea dl at cs.oswego.edu
Thu May 31 20:23:41 UTC 2012

On 05/31/12 12:58, Mike Duigou wrote:
>> So couldn't method hash(Object) be moved to AbstractMap?
> The differences in the avalanche (XOR scrambling) preclude this. It could be
> decided for Java 8 to use a consistent scrambling implementation. I would
> want to hear from Doug Lea whether he thinks this is reasonable/worthwhile.

I think a lot more (mainly empirical) performance analysis is needed
to make this call. But I also think that the forms of JDK8 versions
(and probably backports) are likely to change a lot anyway
(ConcurrentHashMap for sure). So doing this anytime soon is
likely a bad idea.


More information about the core-libs-dev mailing list