Cliff Click

Yes, docs are a contribution! One less thing to think about. Bring 'em on.

Two tests added based on Fridays CCC.  Here's the typing for the original I/K/W term: |  @Test public void test78() {|| ||    rune("I = {x->x};"+|| ||         "K = {x->{y->x}};"+|| ||        ...

Will take a look tomorrow (the other bug was minor, but got social activities distracted so not done yet), but in this case |test| is being as a lambda in...

Heh, nice PR.  Will look at it when the H-M stuff dies down, but honestly seems like the Right Thing To Do. So consider this as an "approved"  AA language...

NBHM is indeed not a wrapper around CHM, but instead implements CM with a very different implementation than CHM. NBHM does not (now) have a 'compute' call directly, so it...

Yes, **impossible** to both be non-blocking and "only compute once" - for all machines, and all implementations of all hashmaps. If 64 cores on a 64-way machine all ask for...

You are not using LockEntry's as keys, instead you are using LockEntry.hashCode as the key. If you have any hashCode collisions from unrelated locks, they can remove the LockEntry you...

:-) Well, if you want some help diagnosing a bug, it pays to hand out all the relevant code...  like, please hand out the entire code (missing pieces in the...

The HashMap contract is pretty clear; the key.hashCode and key.equals() calls cannot change behavior over time.  Other fields in the keys are irrelevant (from a HashMap POV) and can be...

Fixed test cases. Also figured out when the test case checker was failing, because for me on windows it looks like it silently succeeds, when instead it has silently failed....