Each client sends a sequence of mutations to the server, each identified by a mutationID. During downstream sync, server says up to which mutation it has processed.

In Replicache, you can known when a mutation is finalized with the `onSync` API, or more ergonomically, with the upcoming `pendingMutation` API.

https://doc.replicache.dev/api/classes/Replicache#onsync https://doc.replicache.dev/api/classes/Replicache#experiment...

PS- there actually is terminology for this already. Replicache is technically a Causal+ Consistent transactional system: https://jepsen.io/consistency/models/causal