This is a list of items that are planned in future versions of pegasus-mpi-cluster.

* Have workers send task stdout/stderr via I/O forwarding
* Implement a better, object-oriented logging interface
* Allow runtime estimates to be specified in DAG with -r/--runtime
* Implement hard limits on task runtime
* Add BFS and DFS mode via level-based priority?
* Add support for more sophisticated scheduling? (e.g. homogeneous task 
  scheduling like FCFS, EFT, ETF, HLFET, MCP, DCP, MD, and others described 
  in the DCP paper) Some algorithms could use the priority as a strict 
  priority, some could use the priority as a runtime estimate, and some could 
  use priority as a weight.
