At Yieldbot we recently set up MapR with AWS EMR. Consequently, DescribeJobFlowsResponse contains the element SupportedProducts with mapr-m3 in the member field. It seems that since the current boto code doesn't account for the SupportedProducts element explicitly, other elments such as Steps aren't being parsed correctly and returned by the describe_jobflows method similar to issue 460. The code in this change attempts to remedy this by accounting for arbitrary elements Amazon may include in the response; creating a class for them as they encountered during parsing and storing the resulting objects in the instance dict unknownelements of the JobFlow object.
parse arbitrary xml elements and store them in an emr jobflow object