Right now for persisted samples we fetch the entire document from the sample. We could use a query like
find({_id predicate}, {<all of the necessary metafields>, docs.a: 1, docs.b: 1, ...})
but that would make the query not express-eligible.
Now that we are implementing the document chaining approach, it may make sense to apply the projection so that we are not fetching/iterating over irrelevant fields since the document chaining query will not be express-eligible (I think).
- is depended on by
-
SERVER-131922 Refactor sampling code that calculates top level fields necessary for query
-
- Needs Scheduling
-
- is related to
-
SERVER-112233 Generate an appropriate projection for sampling estimators
-
- Closed
-
- related to
-
-
- Closed
-
-
-
- Closed
-