Hi all, (had a nice hike today, and asked the beautiful Lake 22 ;-))
An unconventional patch over an unusual hashcode/equals implementation is a bit too controversial. I'd like to propose a new patch that focus on the problem at hand, that is re-setting the opcode causes the item in HashSet to get lost because of the changed hash bucket. This patch therefore simply remove it from the HashSet before the change and then add it back as if it's a new item after setting the opcode. This patch resolves the issue while leaves BCEL 6.0's hashCode/equals alone.
Here's the new patch: http://cr.openjdk.java.net/~joehw/jdk16/8248348/webrev02/ Thanks, Joe
