Implement Multi-value-register CRDT about antidote HOT 7 CLOSED

The implementation is done and I have added several unit tests to try to make sure that it is properly working. I also manually tested it in console of floppystore and seems it is working.
Please feel free to check and tell me if you agree with the implementation/naming or not. The link is
https://github.com/SyncFree/floppystore/blob/extra_crdt/src/riak_dt_mvreg.erl
I still have two doubts:

1. Where should I put the file? I am putting it in src of floppystore and it can compile with unit test working, but I can not use it when running floppystore. Then I copied it to deps/riak_dt/src and I can use it in floppystore. But it does not seem a good idea to put it in that folder, as it could be easily overwritten...
2. I didn't add any riak_test for it and I am not sure about what is supposed to test. Unit tests seem to be sufficient for it already.

from antidote.

Why did this get merged? According to dialyzer, it looks like this code has quite a few problems:

riak_dt_mvreg.erl:72: Invalid type specification for function riak_dt_mvreg:value/2. The success typing is ('timestamp',[any()]) -> [any()]
riak_dt_mvreg.erl:78: Invalid type specification for function riak_dt_mvreg:update/3. The success typing is ({'assign',_} | {'assign',_,_} | {'propagate',_,_},_,[any()]) -> {'ok',[any()] | {_,_}}
riak_dt_mvreg.erl:100: Call to missing or unexported function riak_dt_vclock:dominate/2
riak_dt_mvreg.erl:132: Call to missing or unexported function riak_dt_vclock:merge/2
riak_dt_mvreg.erl:146: The specification for riak_dt_mvreg:equal/2 states that the function might also return 'false' but the inferred return is 'true'
riak_dt_mvreg.erl:152: Type specification riak_dt_mvreg:stats(mvreg()) -> [{atom(),number()}] is a supertype of the success typing: riak_dt_mvreg:stats(riak_dt_mvreg:mvreg()) -> [{'value_size',non_neg_integer()},...]
riak_dt_mvreg.erl:156: The specification for riak_dt_mvreg:stat/2 states that the function might also return 'undefined' | float() but the inferred return is non_neg_integer()
riak_dt_mvreg.erl:159: The call riak_dt_mvreg:value('timestamp',MVReg::riak_dt_mvreg:mvreg()) contains an opaque term as 2nd argument when terms of different types are expected in these positions

from antidote.

I don't think that this merged?! But maybe I am reading the branch graph wrongly...
My email on the current status says: "Branch mvreg needs check and polishing! But it is not very urgent."

So, your check says: lots of things to fix!
@marsleezm: Can you fix these issues?

from antidote.

I have fixed most of these warning. I pushed it so you can check.

However still there are two left, which are the 161 and 163 line of the code. Dialyzer complains that doing pattern matching to mvreg will break its opaqueness, however I did that for many times in the code and only these two have warning. @cmeiklejohn , can you help?

from antidote.

We need to remove the unrelated QuickCheck code.

from antidote.

I deleted EQC code and made a pull request. Please check it if you have time.

from antidote.

Completed.

from antidote.