-
Type: Bug
-
Resolution: Done
-
Priority: Major - P3
-
None
-
Component/s: None
-
None
Some drivers ignore the 'ns' field in the server's command cursor document for OP_GET_MORE. This hasn't resulted in any bug reports so far because the 'ns' value for an aggregate or parallelCollectionScan cursor is the same as the namespace of the collection running the command.
For example, in python:
>>> c.foo.bar.full_name u'foo.bar' >>> c.foo.command("aggregate", "bar", pipeline=[], cursor={})['cursor']['ns'] u'foo.bar'
However, the namespaces don't match for the new listCollections and listIndexes command. Again, in python:
>>> c.foo.bar.full_name u'foo.bar' >>> c.foo.command("listCollections", 'bar', cursor={})['cursor']['ns'] u'foo.$cmd.listCollections'
The purpose of this ticket is to ensure that all drivers always use the 'ns' value for OP_GET_MORE, regardless of the command.
- depends on
-
RUBY-844 Always use command cursor 'ns' value for OP_GET_MORE
- Closed
-
PYTHON-806 Always use command cursor 'ns' value for OP_GET_MORE
- Closed
-
CXX-455 Ensure command cursors are instantiated with returned namespace.
- Closed