Pedro Teixeira
Pedro Teixeira
@jimpick I think you’re right: we should create deterministic unit tests. I’ll throw something together. @jimpick if we test collaboration/shared.js in isolation, do you think you could come up with...
@jimpick I've scaffolded the test that I think you need in [this WIP PR](https://github.com/peer-base/peer-base/pull/251). Do you want to fill in the deltas and arguments that can cause the convergence to...
@dirkmc thank you so much for starting this discussion! @dirkmc and I bounced some ideas, which are summarised here, but we would like to have everyone's views in here. IMO,...
@mikeal I think the entirety of the delta should be signed, including the link to the parent (to prevent replay attacks). Also, I think we need to represent the collaboration...
@jimpick I love these graphs, they really illustrate well the cases! Yes, I think we can live with multiple computations of the state. A small problem I see with here...
(So the divergent paths here, are not merges, but equivalent paths with different delta granularity.)
Another thing to take into consideration is the delta itself. It's an arbirtrary data structure. The shape of this structure is fixed per CRDT *type*. This type is usually composed...
@marcooliveira could you move this to an google doc so that we could do scoped comments more easily?
@marcooliveira Thank you for this! I've left some comments on the document.
@marcooliveira I think that @jimpick has a point about the offline, I've left some comments there about it also. Also, @jimpick mentioned pinning, and I think that's a (very) interesting...