Persisted struct

[danieleades]

sorry, just a couple of small questions about the Persisted struct.

what's the different between the 'version' and the 'sequence number'? what are these each used for?

Also, I was looking at the inmemory::EventStore and how you would implement #49 on it.
I think if you change the 'backend' from

Arc<RwLock<HashMap<Id, Vec<Persisted<Id, Event>>>>>

to

type Backend<Id, Event> = HashMap<Id, BTreeSet<Persisted<Id, Event>>>;
Arc<RwLock<Backend>>;

then the RangeBounds method are supported pretty naturally. (you would need to implement Ord for Persisted, or use BTreeMap instead).

It's slightly irksome that the Id appears twice in this type signature. Is it possible to move the Id out of the Persisted struct? so that you have

type Backend<Id, Event> = HashMap<Id, BTreeSet<Persisted<Event>>>;
Arc<RwLock<Backend>>

[iitsdani]

what's the different between the 'version' and the 'sequence number'? what are these each used for?

version is the "version" of an Event Stream. Used mainly for optimistic concurrency and avoid data races. In other words, a "version" is the Event Stream-local monotonically increasing number of events in the stream.

sequence_number is the global offset of each event. An Event Store is basically an ordered list of Domain Events. The sequence_number represent the order of these events in the whole list.

As for inmemory::EventStore, and the EventStore implementation in general, we need to rework that one. It's using an outdated design, more on that on an upcoming issue I'll be creating.

As for the type definition:

1. It's not a big deal that the inmemory::EventStore is not efficient, as it should only be used for testing and not production-ready apps,
2. The need for implementing std::ops::RangeBounds is no more ✌️