the problem here is that it is not clear whether High Scale needs a patched ConcurrentHashMap.class to for its library - which I would assume or guessing from fact that they do pack into their jar. so first we need to find out whether this class is there on purpose or not. and maybe find out whether JRuby uses this feature which uses this ConcurrentHashMap

but in the long run I agree that jruby should not pack such classes, other places then ruboto will run into the problem sooner or later.

wow - quite an intrusive surprise. for the record the high scale library was introduced only to use a concurrent IntHashMap alternative: NonBlockingHashMapLong ... which does not depend on ConcurrentHashMap directly.

from the README it seems the patched classes are optional (and thus not really needed to be packaged by JRuby): https://github.com/boundary/high-scale-lib#original-readme
p.s. they also have a patched java.util.Hashtable class

this is now also resolved with #3560 (will be merged over to master)

Thanks!