I don't understand. This page doesn't seem to actually explain how the "merge nodes" actually works to, well, merge divergent histories.
Does it rely on some application-level logic implemented by the user? If so, how is this useful? Existing tools like y.js seem to be vastly superior in terms of usability.
No all the logic is in the library. They are merge edges (not nodes). Merge edges help us to establish the order in which users make changes to the data. Conflicts (divergence) are fixed by the algorithm looking for edges that might conflict. Determining the winner of the conflicts and then replaying the graph
3 comments
[ 4.4 ms ] story [ 16.8 ms ] threadDoes it rely on some application-level logic implemented by the user? If so, how is this useful? Existing tools like y.js seem to be vastly superior in terms of usability.
How does it "determine the winner"? This is the hard part of CRDTs, and you still have not provided any actual details.