prep for cutting a release, target 2019-01-13

[Hal Murray]

>> I think what's going on is that only the best samples make it through to 
>> peerstats where best is lowest RTT of the last 8 samples.  Lowest RTT means 
>> lowest queuing delays in the network.
> Possible, but shouldn't then the sample be even further apart?  or is it just
> dropping outliers? 

There is no concept of outliers at that level.

Each response that gets past the low level filters gets shifted into the 
buffer of last 8 samples.  If it is the new min RTT, it gets passed to the 
next layer of processing and logged via peerstats.

-- 
These are my opinions.  I hate spam.