Users should be able to

• set up a simple pipeline (probably without specifying files),
  • provide a directory with files in it
  • run the pipeline on the files in that dir and put it in another dir

One way to do it is to do an implementation of the Replica Catalog interface in Pegasus that does the directory listing on the input directory, and creates a Replica Catalog in memory of the basis of the list of files in that directory.

• The DAX remains abstract ( no physical paths specified for input files).
• We will have to tinker with the Replica Factory also , as we may still want to register the output files in some other replica catalog