Hash#rehash does not properly rehash

[ChrisBr]

Environment

• jruby 9.3.0.0-SNAPSHOT (2.5.0) 2018-01-08 4b5cea6 OpenJDK 64-Bit Server VM 25.151-b12 on 1.8.0_151-b12 +jit [linux-x86_64]

Expected Behavior

When rehashing a hash with equal objects in it, they don't get removed.

a = [1,2]
b = [1]
h = {}
h[a] = true
h[b] = true
b << 2
puts h.rehash # { [1,2]: true, [1,2]: true } 

I already looked into the issue and it seems that the problem is that Hash#rehash does not handle collisions properly. When two objects have the same hash / bucket number, they get inserted even if they are the same (see example).

Actual Behavior

Hash#rehash should remove the duplicate entries.

[headius]

I looked at your impl, and it was fine except for a couple things:

• entry.equals checks equality of values too, which we don't want
• we need to remove not-added entries from the insertion-order chain; if an entry is not kept, prevAdded and nextAdded entries need to be fixed up
• we need to check the whole chain of entries, not just the first one

I almost had a fix but I'm pulled away to finish something else.