Rewrite the dataset directory service.

The changes are:
1. Use a class for metadata instead of keeping it in pairs.
2. Store the serialized record descriptor received per job in the dataset
   directory records.
3. Implement a method to return the serialized record descriptor when
   requested by the client.

git-svn-id: https://hyracks.googlecode.com/svn/branches/fullstack_hyracks_result_distribution@2830 123451ca-8445-de46-9d55-352943316053
1 file changed
tree: 5c900a5844a35406fa4e8a39970063741fb2967c
  1. algebricks/
  2. hivesterix/
  3. hyracks/
  4. pom.xml
  5. pregelix/