I didn't look into it very deeply, my first impressions are that it looks good but I don't think the idea should be pursued because NIP-34 and Grasp are immediately compatible with other git tooling that already exists and that is a huge plus.
Also hashtree seems to be much harder to implement and I'm a believer in simpler solutions and having dozens of implementations.
Hashtree is very simple, it's just blossom + "this is how you define directories and chunk files". Merkle roots are saved in nostr events so you can have npub/path directories whose content you can edit. Files.iris.to supports nip-34 pull requests and issues. You could use it as one remote even if you also use grasp.
That's good to know. I guess we can have NIP-34 and two remote types: grasp and hashtree living together them. I hadn't thought about it that way.
