Hi everyone, I implemented _bulk_get support for the replicator in https://github.com/apache/couchdb/pull/4144. I tried to make it as backwards compatible as possible to replicate with older endpoints and PouchDB. Some preliminary benchmarks showed a decent speedup of 30% or so.
There is a silly chart in the PR comments showing the effect of the exponential moving average to make it less mysterious where some of those constants came from. I had opted to make some of those constants hard coded, to avoid adding even more replication options, but can be convinced otherwise. Feel free to suggest or comment in the PR. Also, to add some visibility regarding _bulk_get effectiveness I added two replication stats fields. Not sure I picked the best names, let me know if you have any better suggestions or perhaps not have them at all? Thanks, -Nick
