Now, we can use object storage (MinIO, S3, R2 or anything you like) for synchronising! Moreover, despite that, we can use all the features as if we were using CouchDB.
Note: As this is a pretty experimental feature, hence we have some limitations.
- This is built on the append-only architecture. It will not shrink used storage if we do not perform a rebuild.
- A bit fragile. However, our version x.yy.0 is always so.
- When the first synchronisation, the entire history to date is transferred. For this reason, it is preferable to do this under the WiFi network.
- Do not worry, from the second synchronisation, we always transfer only differences.
I hope this feature empowers users to maintain independence and self-host their data, offering an alternative for those who prefer to manage their own storage solutions and avoid being stuck on the right side of a sudden change in business model.
Of course, I use Self-hosted MinIO for testing and recommend this. It is for the same reason as using CouchDB. -- open, controllable, auditable and indeed already audited by numerous eyes.
Let me write one more acknowledgement.
I have a lot of respect for that plugin, even though it is sometimes treated as if it is a competitor, remotely-save. I think it is a great architecture that embodies a different approach to my approach of recreating history. This time, with all due respect, I have used some of its code as a reference.
Hooray for open source, and generous licences, and the sharing of knowledge by experts.
