Java Maps do not support rehash, so that's why we raise NotImplementedError. I suppose we could just do nothing, since rehash is largely an invisible operation.

For Java arrays, shift has no meaning; they are fixed-size. I think in this case we need to preserve the NotImpl error.

For Java arrays, shift has no meaning; they are fixed-size. I think in this case we need to preserve the NotImpl error.

I mean we should leave it unimplemented. Java arrays are not like Ruby arrays. If you want a Ruby array, call to_a.

May I ask why you are calling these methods against Java maps? Perhaps we can help you find an alternative, because at least in the case of shift I don't think we can do anything.

I should point out that rehash does work when the particular Java Map supports it, but it doesn't exist for all Map impls:

[] ~/projects/jruby $ jruby -e "java.util.Hashtable.new.rehash"

[] ~/projects/jruby $ jruby -e "java.util.HashMap.new.rehash"
NotImplementedError: rehash method is not implemented in a Java Map backed object
  rehash at org/jruby/java/proxies/MapJavaProxy.java:310
   <top> at -e:1

Thx for the reply.

I'm writing a custom plugin for logstash.
The events that I get when my code is called by the main application are Java maps.

If I use to_a, I'm afraid I will get a copy of the event, and won't be able to modify referenced map itself.

Re maps: I can make rehash work, but there's no equivalent to shift for Java maps.

Re arrays: Java arrays have fixed size, so size-altering methods will never have an equivalent. Not really anything we can do about that :-\

I've pushed the rehash change, and improved the shift error message. That's about all we can do, so I'm marking this fixed.

Ok, thx anyway!

Oh, one last change I will do...if I mark the default rehash and shift as unimplemented, they should not show up in respond_to? checks.