Implement output option for aggregation pipelines only to unsharded collections. Input collections can be sharded.
This should be considered a part of SERVER-3253. SERVER-3253 dealt with commits that only dealt with unsharded clusters.
sharded map-reduce output should be parallelized and properly distribute chunks
support cursor based aggregation
$tee command for aggregation framework
Aggregate Command will support $out
Support $out aggregation pipeline operator
new aggregation framework
Message: ExpressionContext now has the input namespace
This reduces the amount of PipelineD injection and will make sharded
$out easier (SERVER-10097).
Message: Make doingMerge a property of DocumentSourceGroup rather than ExpressionContext
Prep for sharded $out (SERVER-10097). Will need to be able to send a
merging $group to a shard.
Message: SERVER-10097 sharded $out
Note that all sharded aggregation tests now go through the
"merge-on-mongod" codepath now rather than the "merge-in-mongos"
Message: SERVER-10097 Visual Studio – add db/pipeline/document_source_merge_cursors.cpp