Active revision index

[pmelab]

For a couple of use cases (finding the current active revision, comparing revision between workspaces) it would be beneficial to have an index of all "active" revisions for a given set of contexts.

Assuming we have two context levels K and L

       A
     /   \
    B     C
  /       | \
 F        |  D
          | /
          E

On creating a new revision, this list could be materialized into an index table for each entity type:

entity_id	revision_id	context_k	context_l
1	A
1	B	X
1	E	Y
1	E	Y	Z
1	F	X	Z

This table could be used to performantly query for the active revision, which would require a couple of drastic joins and subqueries right now. Also it would allow query operations between contexts (workspaces) like:

• "Retrieve all nodes that contain updates when moving from workspace A to workspace B"
• "Retrieve all nodes that have leaves in both A and B and therefore are potential conflicts."

In this scenarios workspace A and B are not necessarily related.

Edit: To enable the latter case, we will also have to maintain a table of all parent relations.

[pmelab]

Thinking more about this led me to the conclusion that we would anyway require an index table of ancestors and a sparse tree of "active revisions" would be a performance enhancement.