Closed jrwest closed 10 years ago
This is probably to late for 2.0 in general. We do need to revisit if we want to continue to use these clocks long term. Marked for 2.0.1
but may slip to 2.1
My opinion is that, since DVVSets are used as the model for the riak_object enhanced causality tracking, they've been quickchecked.
Aren't they different in riak_object (our dots impl vs the original dvvset.erl used by cluster meta)?
Sent from my iPhone
On Mar 25, 2014, at 1:24 AM, Russell Brown notifications@github.com wrote:
My opinion is that, since DVVSets are used as the model for the riak_object enhanced causality tracking, they've been quickchecked.
— Reply to this email directly or view it on GitHub.
Sure, different, but similar enough that one is the model for the other. Pretty sure that the model and the thing it models get tested in equal measure with EQC.
thanks, in that case I believe we can close. and for completeness, the model being referred to is https://github.com/basho/riak_kv/blob/develop/test/riak_object_dvv_statem.erl
As mentioned during review of basho/riak_core#363 the included
dvvset
module was only given a cursory review. A decision on whether to continue using it or to convert it to using plain old vclocks and further review ofdvvset.erl
if it is chosen is necessary.