[Auto|Hyper]merge

[alfl]

Interested in thoughts on the Hypermerge project from the Automerge team.

As we think about using CRDTs to merge data, are these projects useful to us? Are other projects better?

Does something like this make sense wrt n-quads? Websockets to communicate in real-time between cores?

[maurerbot]

In a lot of use cases, peering with a remote core would help resolve order conflicts. This isn't great for near real time use cases.

In that case, a general crdt opinion might be useful but, we should expose the a means of developers to BYO toolkit.

[szkl]

we should expose the a means of developers to BYO toolkit

I agree. There should be ready-to-use components/modules but also define abstractions and provide APIs so it is extendable. I mean we should have this approach in general.

[Cosmin-Parvulescu]

I wonder how realistic would be to use Hypercore and perhaps Hypermerge to actually solve the sub-core distribution issue, @szkl

[szkl]

I need to read about them first :)

[maurerbot]

After several experiments with Automerge, Hypermerge, and YJS we're moving forward with a more sumple object versioning strategy. CRDTs will be more useful for "chat room" like use cases in the future.