Introduce optional task/data object groups and names for better monitoring and possibly debugging. Currently, the data objects have a label indicating the role it plays in the producer task (e.g. log, score, ...).

Introduce:

• task and data object name - arbitrary string, optional, to be set for any tasks the user wants to distinguish later by their name. May not be unique in the session but is intended to be unique. Always set by the user.

  • Goal: be able to distinguish special tasks in the graph and to query them.
  • Importance: low
  • Perhaps we need to think about possible uses and name vs label on objects (not to be confusing)

• task and data object groups - a textual label splitting the nodes into disjoint groups (including the None group).

  • Goal: to be able to monitor statistics of different task/DO groups, meaningful progress reporting and stats aggregation
  • Advantage: easy to report compact per-group aggregates, even store them in the log.
  • Variant: add arbitrary tags, aggregate by tags combination. Drawbacks are added complexity (to many tag combinations) and less monitoring meaning (the stats no longer add to all nodes).