If anything, the DB version will make users further dependent upon commercial sync options, as only heavy-clients (desktop, native apps) will be able to synchronize database state. There is no reason to think that having a database as a persistent store will entail or prompt the creation of an API layer. All of the calls to the database can be internalized in back-end Clojure code.
I’m building a browser-based, server-hosted (self-hosting), open source, markdown-based, multi-repository, non-commercializable (MIT license + Commons Clause seems to be the best solution for that, even if that is not strictly FREE according to FSF), outliner/note-taker that backs up to github (or any remote configured in the notes repository). There is no collaboration, nor will there be, and no dependency on sync solutions. Not currently multiuser but I’m working on that. Encryption-at-rest is baked in via age encryption. Mobile interface, PWA is in progress. Working on intelligent sync conflict resolution. There is no AI functionality, but if I expose extension support I’m sure someone could build it. The application is built with the assistance of Claude Code, which, yes, I know, but for rapid prototyping it’s a godsend.
I’m a long, long way from prod and the code is almost certainly woefully inefficient right now, but I wanted to get in place what I want and work on making it how I want it later. I’ll drop a link here when I’m not embarrassed to call it v.0.1. I have no intention of ever monetizing this. If it goes somewhere and is useful, it is a community resource for the good of all.
There is a self-host sync option with Logseq DB, for those wanting to tinker …
You can follow along in this thread Discord
