[gbadman]

This is a really cool and ambitious project. Although some mention ShareJS, I would say that the OT portion more closely resembles that employed by Firebase in their Firepad project [1] (in that no centralized server manages ordering and history). Actually, this system has many of the features of Firebase in a self-hosted package.

As a user of ShareJS (for the real-time collaboration on Plunker [2]), I can see some interesting benefits offered by TogetherJS; the system has cursor and presence features out of the box.

That being said, the main drawback that I see in TogetherJS is that it seems to be tightly coupled in a way that it is hard to take smaller parts of the overall system. It would be great if the P2P network system could be extracted and plugged into a node.js Stream interface. That would allow the code to plug into a whole ecosystem of interesting libraries [3].

I'm really excited by all the powerful libraries that Mozilla has been pumping out these days. The web is a great place to be.

[1] https://github.com/firebase/firepad

[2] http://plnkr.co/edit/?p=streamer&s=TogetherJS (Demo)

[3] https://github.com/substack/stream-handbook#introduction

[ianb]

[Author here] The big way TogetherJS differs from many of the other realtime products out there, like Firebase or (to a lesser degree) ShareJS, is that it doesn't do any storage. It coordinates real-time sessions, but it's not there to take the place of the application, instead it augments the application. If your application has persistence, it has traditional persistence. Whatever auth system you have is your own. At the same time while it dodges any persistence, it extends the scope beyond what those other tools typically do to include more with session management and the standard tools that I think are needed for collaboration. There's nothing exciting about text chat, but you need it anyway.

Still, ignoring persistence I think will help keep the surface area of the TogetherJS smaller compared to some other tools. Or at least focused on other things.

People have mentioned the idea of being able to chop TogetherJS up before, but I have a hard time seeing how that would actually work. For instance, the architecture is built around a pretty dumb server, so hooking into Streams would seem counter to that principle. And just allowing for the possibility would add a lot of complexity. Ultimately the UI is actually the bulk of what makes up the TogetherJS code (or maybe it's the part I find hard and so I feel like I spend a lot of time with it), and I don't know how to make this kind of UI while being modular. Without the UI I don't think it's nearly as compelling a project (becomes more like ShareJS, or Firebase, or Google Drive Realtime API – neat but a big investment, and sometimes only appropriate for greenfield projects.

Anyway, I'm certainly interested in how people want to extend or integrate with TogetherJS. I'm skeptical of splitting it up into pieces, as it is theoretically easier to do things that way, but I expect in practice it will just leave a confusing pile of pieces that aren't fun to put back together. But that doesn't mean that there aren't use cases worth exploring... and with some creativity there might be other approaches to handling those cases.

[KaoruAoiShiho]

Is this a free pusher? What's the catch?

[davidascher]

[mozilla employee here] There's no catch except an absence of a commitment to keep running it. Mozilla Labs is charged with figuring out how to make the web better. We think this could help make more apps collaborative. You tell us if we're right.

[mbreese]

It doesn't seem to be a P2P system, since it requires a server. According to the docs, "This server does not rewrite the messages or do much of anything besides pass the messages between the participants" [1] , so it's not really P2P.

I'm fairly sure the Firebase requires a server too... (their server, or else they couldn't charge for it).

[1] https://togetherjs.com/docs/#technology-overview

[ianb]

On principle we keep the server simple, echoing messages back and forth between clients. In practice it's quite hard for two browsers to connect directly to each other, and while we want to support that kind of architecture in order to get it working we still need to simulate P2P using a server.

[derefr]

> In practice it's quite hard for two browsers to connect directly to each other

Isn't this the problem WebRTC abstracts away? Your server should just be a TURN server (http://en.wikipedia.org/wiki/Traversal_Using_Relays_around_N...), shouldn't it?

[ianb]

WebRTC makes it possible, but not easy, nor abstracted. You still have very long connection strings to establish the connection, and those strings change as you move around, so it's not something we'd want to ask users to send to each other in order to establish a connection. The approach usually used is a server where the two people rendezvous to exchange the connection strings. That server is kind of like what we have – but then once it's in place, changing over to WebRTC just doesn't seem as exciting. (And even then the connections are slower to establish and less reliable than WebSockets.)

[k__]

AFAIK WebRTC just works with the latest Firefox/Chrome and needs a STUN and/or a TURN server.

[deliminator]

We provide wysiwyg editor integration (Aloha-Editor) with presence and remote cursor. We are working on a hosted/downloadable service with low-level data-sync api and high level wiki features: https://wikidocs.com

[meowface]

Does anyone know of something like this, but for Markdown?

So, like a collaborative Markdown document editor, as well as a simple way to store those documents and choose them for reading/editing later.

[gbadman]

You can actually do that in Plunker right now by having an `index.md` file instead of `index.html` [1].

In the absence of an html index, the plunker preview server will try transpiling to html from different languages. Markdown is currently supported.

Also transpiled are things like less, scss, coffee-script and more. Essentially, your html requests a file with the target suffix (e.g.: .css) and the preview server will search for compatible source files to transpile to that target.

[1] http://plnkr.co/edit/?p=streamer&s=markdown

[mburns]

https://github.com/JohnMcLear/ep_markdown

[rspeer]

Wait, that's backwards. The point of Markdown is that you type the formatting quickly, instead of clicking buttons on a toolbar.

That animation seems to be about clicking buttons on a toolbar that will secretly generate Markdown formatting for you.

[jamest]

http://socrates.io, built with Firebase.

[leephillips]

I agree that it's cool and ambitious, but it seems to have a large intersection with Google Wave.

[jeswin]

It integrates with your existing UI, and not as a separate box like Wave or other plugins. I also like that this is free, open source and _from Mozilla_.

[leephillips]

All true, but note that the Wave server is free and open source: http://incubator.apache.org/wave/about.html