ResultWriterOperator should provide the record descriptor of its input buffers at run time to serializer and it shouldn't be statically provided to the factory provider at static query compilation time.

This is because the final output of the results being generated may be different than the
statically known record descriptor in some queries like aggregate queries, so we need
to dynamically provide the record descriptor at runtime.

git-svn-id: https://hyracks.googlecode.com/svn/branches/fullstack_hyracks_result_distribution@3055 123451ca-8445-de46-9d55-352943316053
3 files changed
tree: 80f66d614af252502aa53b6dcaa94a237cc8455f
  1. algebricks/
  2. hivesterix/
  3. hyracks/
  4. pom.xml
  5. pregelix/