Kevin Jahns

Thanks for the repository! I fiddled around with it a bit. I don't think 0.2.88 broke jest. It probably broke when `webcrypto` was introduced to lib0. `lib0` makes a lot...

Hi @josephg, Thanks for offering to share! How do you currently store the editing trace? For a linear editing trace (no concurrency), I suggest expressing it as an array of...

The deleteset is pretty small. So I suggest exchanging it with all connected peers. My reasoning: Comparing snapshots (which contain a deleteset) is less efficient than exchanging state vectors and...

> It's more efficient when you can know if a given remote peer's data is a superset of yours (i.e. if merging your snapshot into theirs doesn't change it) even...

> This does not work currently. > e.g. Simple text CRDT, delete one line, nothing else, results in: state vector totally unchanged. There is a difference between state vector and...

There is no definite answer to this. And it really depends on the application / UI framework you are using. Bartosz can correct me, but I think that a Shared...

Looks good to me! Ideally, we would also get rid of `get_text` then, which is an awkward concept. Maybe `txt.get('text')` or even better `txt.root.text`. But I'm not sure if this...

Moving objects between different types has significant drawbacks. We would only be able to do limited garbage-collection, and most users would find the conflict-resolution behavior hard to comprehend. It's not...

> Apparently not all updates are guaranteed to be stored (is it itself a bug? This needs to be documented at least). For example, if you do an update and...

All updates are stored immediately. The mentioned timeout is only for trimming the updates. Persisting can fail if you close the web-app before the last transaction finishes - this should...