Return whole nodes when querying existing partitions

[enikao]

In simple bulk API, we have an operation to retrieve all existing partitions. What shall that operation return?

Partitions are always nodes (#29).

Variant A: Only node ids of partitions

Return only a list of node ids.

Pro:

• No overhead

Con:

• There is very little we can do with only the id. We don't know the concept (or any feature), so we cannot decide whether this partition is interesting to us on a technical level. So we almost always need to query again to retrieve details about the partitions.
• New serialization format in the protocol

Variant B: Complete nodes of partitions

Return a SerializationChunk with the complete partition nodes, but no children/annotations.

Pro:

• Easy to process technically, as we know concept and features
• Can be shown as a selection to the user, as we might get a name property

Con:

• Potentially lots of information we don't need

Variant C: Complete subtrees of partitions

Effectively return the complete repository.

Pro:

• Get everything at once

Con:

• No way to reduce overhead

Decision: Variant B: return complete nodes of partitions, but no children/annotations.

[joswarmer]

Should close this