I know that being clever with the aggregate transform can do many useful powerful things but it might be nice to have the ability to merge two result sets together. The Merge would potentially allow multiple input sources. When the internal logic sees a Merge it can stop and trace back through the lineage to the beginning of a merge chain.  This would require that an input source be listed as a primary so that you always have a 'starting point'.

e.g.:

    XML reader (primary) -> some transform -> another transform -> MERGE

	XML reader (primary) -> some transform -> another transform -> MERGE

    CSV reader (secondary) ----------------------------------------^

	CSV reader (secondary) ----------------------------------------^

This would supplement the Aggregate transform.  Obviously the number of fields would need to match and hierarchy would be no fun to deal with but I think there's good value there.