Chat room.

gistfile1.md

So, we figured creating our own chatroom could be fun.

It's something we can all work together on.

We don't plan on moving there but the project itself could be interesting for the whole room to collaborate work on.

What we need to discuss now:

• What technologies we want to use (Node/Express seems natural for serverside. What about clientside?)
• What features to we want? Do we just want to start with a clone of the SO chat?

Requirements

• Adding features should be easy
• Responsive and fast
• Works on IE10+, Chrome, Safari, Firefox, Opera
• Mobile interface that actually works

Current stack proposal

• AngularJS as clientside MVW, libraries (like jQuery) accepted gladly on a need to basis.
• NodeJS/ExpressJS/SQL for server side.

Actually, this is about fun. We can ditch jQuery if it makes us feel better about ourselves. Angular comes with jqLite anyway.

We can do lots of prollyfilling too.

yup. if this is meant to be an exercise and something fun and challenging why not shoot for the "as few third party dependencies as possible" stars. I would love to see this done with no or a minimal css framework (purecss.io is nice) and no or minimal DOM abstractions. It would be nice to have an // authored by: <username> under each script that was actually authored by a room owner / regular

@benjamingr Who is starting this empty repo to start with @rlemon ?

It goes without saying that there should be absolutely no jQuery.

Aside from that, we should write it in brainfuck.

https.

I don't think we should go with a framework server side, the REST API should be quite small.

Postgresql of course, it's a no brainer for the transcript. Maybe a redis frontend for pub/sub would be nice, but I think I'm over engineering.

Client side I'd go with angular + bootstrap because I'm lazy and it does what we need it to do.

On the architecture side, going full events is probably the best way since it will allow us to easily plugin new features. And it's familiar to everyone. Events should propagate objects that are modifiable, and promises with bluebird will be used whenever necessary to avoid callback hell.

Welcome.

There's a SOJSChat organization on GitHub, this should be put under that.

One of us should be PM of the whole thing. I've seen too many community efforts fail for lack of a central organizer.

I hear cats are natural organizers. The internet probably has something to say about that. Are you volunteering?

@JimmyHoffa We're very good at what we do.

I can PM, but that means I wouldn't be able to write much code, time-wise. Pretty busy right now.

The following URL now forwards to this gist:

http://new.javascriptroom.com

+1 Angular
+1 MongoDB (or CouchDB)

+1 no jQuery
+1 Mongo or Couch
+1 Angular.

Private Messaging is a major feature I would love to see.

+1 for no client-side MV*, no jQuery, no Mongo, no Redis, no Node, no JavaScript.

… um.

I am willing to contribute www.lobby.ws to the cause. I don't care for the ws tld, but I am fond of this name for a chat room service.

i have chats.ws iirc . lemme check though xD

Taking florian's suggestions and Benji's recommendations and our choice of UI I designed something last night here is the first screenshot of it. http://i.stack.imgur.com/ivetb.png

Added to that i suggest a mute and a Warn feature

a warn displays a dialog on the targetted screen telling him to something in bold ( this feature will be owner only )

a mute just mutes the user in that chatroom for x amount of time if x = -1 , thats a forever mute

Nice, this looks good.

As for no-jQuery, I say: +1. Also, f* IE8. I want Opera Mobile support, though.

As for writing the core: not sure I'm the best to design an architecture. I have that on my resume, but I don't want to do that singlehanded.

Honestly mongo is a no-go for me. It doesn't do anything better than postgres. Transcript is not something that will change its structure, so going noSQL isn't really useful.

nice :)

most of the important stuff has already been said, from my side:
+1 for Angular and basically the setup Ivo Wetzel suggested.

I don't care too much about using or not using jQuery, with the given browser-list that should be supported we won't really need it.

Features:
private messaging, multiple rooms and nice code-highlighting when posting code-snipplets would be great

Cannot forget about onebox support.

If we're really doing this, why not add WebRTC support? Make it more like a hangout?

It’s definitely a good idea to show off as much JavaScript as possible, so I’m all for that! I don’t think it’ll see much regular use, though.

Several ideas...

1. A user preference to stop autoplaying of GIF's, that will show the GIF as a static image with a play button over it.

2. A library of audible emotes with a user preference of whether or not to auto play the emotes or not, because sound clips can be funny at the right moments, but also annoying at the wrong moments.

3. Third is more of a principle:

I think what would really add a lot of quality to a chat service, is to have subtle user preferences for anything UI related. This may include things like:

• disabling oneboxing by the site
• themes
• choosing a markdown flavor
• whether or not messages are prepended or appended to the message list
• if user list goes on the left or the right
• other layout related things like that

The principle is this; User preference all the things!
The obvious problem is finding a way to pull these things off without bloating the UI.

Which is why we would need to consider this early in the development.

Considering the "We don't plan on moving there" part, maybe it should be built on top of the SO chat. Then anyone could use all the new chat features/UI without needing to leave the SO chat.

I have no clue how doable that is with the SO API though.

Or maybe that's already the plan?

This is what I recommend

Server:
Mongo (possibly Mongoose)
Express
Socket.io
Client:
Javascript
socket.io
reset.css

Well, it looks like I'm majorly outvoted, but I happen to like knockout as an alternative to angular.

Firebase also looks like an interesting option, though I admit I haven't actually used it.

put all this on s3 + heroku ... infinite scalability :D

@shea i once wrote a lib to convert a gif to a sprite sheet ( in C ) I think i should do that :P ... making the gif pausable and playable with canvas or css3 powered animations !

@benjamingr any updates? ;-)