Aggregator requirements and wish list:

The Aggregator is the class, which applies changes to the data. it is still unclear how this interfaces or should be integrated in the harvesters. For now it is disconnected, but most metadata information can only be known by the harvesters at the time of harvesting, so it probably be closer connected.

• integration with harvesters
• There are aggregations, which cannot give back a patch, since they have to be done before data is parsed to a graph structure? for now we log in the patch system what was executed, but we did not went into making a graph or so. since we are still unsure about the whole architecture, what to do with graphs and what not.

For aggregating data on disk or for testing a cli exposure is useful

• expose on cli: hmc-unhide aggregate