On 29/05/2013 15:28, Peter Levart wrote:
.....
I don't feel strongly about this either, but I think it deserves
possibly its own bug number and consideration. I have removed it from
this review request, and will a file a new bug to track it.
Hi,
Why not using Unsafe (which is already used in CHM) to re-use the
AbstractMap.keySet/values fields? They could even be accessed with
normal non-volatile read/write although they are declared volatile in
AbstractMap. Is this to "hacky"?
Possibly a little hacky, and I guess may sacrifice a little performance?
But this kind of analysis/investigation, and any compatibility concerns,
are exactly why I think this deserves its own bug. I'm not saying that
it's not worth pursuing, just that we should decouple it from the other
changes going on here. (Says me who wanted to resync, in one pass, the
complete j.u.c last week!)
-Chris.
Regards, Peter
