Fork me on GitHub

There's like this project: Maybe it can be useful, at least as an inspiration?

👍 7

So... by virtue of datahike being implemented via a persistent data structure (the hitchiker tree), it's already possible to do things like fork and clone (with structural sharing)., So if that's your interest, we're already there. Merging is the real kicker, and frankly I'm a little skeptical about in relation to dolt, given the scant documentation about it in the README. Conflict free merging implies a CRDT structure, which corresponds with constraints around the kinds of things you can express in the data-language, and imparts a heavy toll in terms of how you model things. Dolt seems too "easy" for me to expect that it would always have the desired results. That having been said, @whilo has thought a lot about this and how deal with merging and resolving conflicts (e.g., and I think it's an ultimate goal to incorporate this kind of functionality into datahike (though lower priority presently than things like performance, datomic parity, front-end support, etc).


So is it possible to implement datahike on top of git? Maybe; maybe even probably (though I think the question is with how much work). The real question is what would it get you, and I think the answer is not much given that the project is already headed towards some of the same potential value propositions in more principled (and likely performant) ways.


I could obviously be misunderstanding any number of things here, so please let me know if that's the case, but hopefully this helps a bit.