[Proposal] Add support for entity persistence & snapshots

[laertispappas]

Hello everyone and thank you for yourk work with RES.

I would like to raise a question and add a small proposal on the data layer which I think it would add some improvements.

I'm facing issues to understand the role of streams in the modeling and why each event emitted should go to a specific stream. I find it more confusing when I have to also specify the stream for each entity (Order$ID).

What I also I think it would make sense is to revisit the persistence layer a bit and add something similar to:

table: events (persists events for a given entity)
event_id | event_type   | entity_type | entity_id | event_data | ...


table: entities (one row per entitiy)
entity_type | entity_id | entity_version (optimistic locking) | ...


table: snapshots (would be great to add support for snapshots)
entity_type | entity_id | entity_version | snapshot_type | snapshot_json | triggering_events |

I think the above will add a couple of advantages to the modeling because of the following:

• Aggregate types and Ids are stored together with events to declare for which aggregate the event is emitted. Which is the case in event sourc
• This is needed because as is now the clients are forced to pass aggregate type and ids to each event which adds some redundant code around + mistakes in case you forget and you expect to parse it later on.
• Assuming one stream per aggregate the above will also take streams out of the equation for callers that don't need to care about streams, but rather persist events for a specific aggregate -- which is the case most of the time.

The entities table is mostly used for optimistick locking. Last but not least it would be great to add support for snapshots. Any feedback much appreciated. Thank you in advance.