automerge VS FluidFramework

Yes. I asked the devs about ideas for this in this ticket and got an interesting response. It's aimed towards server-side handling, but the same ideas apply to local storage as well.

There are a lot of implementations of CRDTs out there. In JavaScript, for instance, we have Y.js (https://github.com/yjs/yjs) and automerge (https://github.com/automerge/automerge). There’s also a Y.js demo (https://demos.yjs.dev/prosemirror/prosemirror.html) that allows you to play around with them and have your own collaborative app running in just a few seconds. All messages are exchange via webRTC and manages the state via CRDTs. This can be a great sandbox to understand how CRDTs work and see.

Anyone unsure of what a CRDT is, this is the perfect intro: https://www.inkandswitch.com/peritext/

The two most widely used CRDT implementations (combining JSON like general purpose types and rich text editing types) are:

- Automerge https://github.com/automerge/automerge

- Yjs https://github.com/yjs/yjs

For instance, practicing "local first web" using automerge with all it's Distributed Persistence Primitives on CRDT's and Vector Clocks - i.e. when the Browser View is treated like a Database replica, essentially; or adopting a real data mapper that's giving you an API from your database Schema, using Prisma or Hasura... or even implementing a custom codegenereted one, as a babel plugin, on top of TSED and Micro-ORM.

And about conflict --as already mentioned in comments-- CRDTs is the way to go here. Automerge is an implementation you may want to look into.

Have you seen FluidFramework? It's open source (MIT): https://github.com/microsoft/FluidFramework

I think the first product they're building on it is Loop: https://www.zdnet.com/article/microsoft-introduces-loop-a-ne...

The Office 365 backend uses git to store snapshots of documents. [1]

https://github.com/microsoft/FluidFramework/tree/main/server...

Full disclosure I work at MSFT and on the fluid framework.

If you are interested in this you may also be interested in the fluid framework, https://github.com/microsoft/FluidFramework

We use websockets and solve a lot of the state management problem called out here by keeping very little state on the server itself. The primary thing on server is a monotonically increasing integer we use to stamp messages, this gives us total order broadcast which we then build upon: https://en.m.wikipedia.org/wiki/Atomic_broadcast

Here are some code pointers if you want to take a look:

The map package is a decent place to look for how we leverage total order broadcast to keep clients in sync in our distributed data structures:

(Disclosure: Work at Microsoft, but I work in Azure and some open source stuff, not on or directly with Fluid/Office/etc.)

That's just a trademark clause for Microsoft logos and brands. The Fluid Framework itself is [MIT licensed](https://github.com/microsoft/FluidFramework/blob/main/LICENS...) and doesn't require exposing any of those logos/brands when you use it, so the framework itself is fairly open for usage.

I think the main thing that would slow down adoption for Fluid is that the only "production" backend is an Azure service, which isn't part of the open source Fluid Framework. [Other open source backends](https://fluidframework.com/docs/deployment/service-options/) aren't recommended for productions. Until there are some open source ones, I'd assume adoption will be limited to folks in the Azure ecosystem.

Which looks like Quip, which looks like OneNote, which looks like a million other note-taking apps before it.

I think the more interesting aspect is the extreme coauthoring; it looks like the underlying tech is open source as well which is pretty interesting for a lot of applications: https://github.com/microsoft/FluidFramework

Within the context of the Microsoft-verse, Fluid Framework (https://fluidframework.com) is supposed to be solving similar problems in web apps, although I haven't personally played with it.

Microsoft recently open sourced fluid framework. It is a distributed, consensus based, real time collaboration framework written in typescript. Fluid would keep your clients synced up and your server code would only have to handle when someone hits submit. Fluid Framework

The official documentation and the github repository seem clear.