ashwinraghav

Its 12 billion light years away. Just keep in mind that this is what it looked like 12 billion years ago => it takes the light 12 billion years to reach the Earth. So the image as seen is really that old!

Firstly this kind of vulnerability requires no white-boarding. It is fair to assume that a dev with a reasonable understanding of how the web protocols work will not need a white board to figure this out. So firstly the argument of catching this on a WB is a little juvenile.

Associating TDD to being haphazard is pretty random. Being hap hazard and TDDing are really in 2 different math spaces. Care not to mix em?

Ah well, I guess there is nothing wrong in getting back to an 'old' yet 'proven' solution.

Here is the ruby library that I used https://github.com/ashwinraghav/marvin. It can help you write a pure ruby application that you can later convert to an event-driven app(If you wish)

Thanks for bringing up the web-app vs chat bot straight away. I did get some questions on these lines during the conference itself. 1) I do not agree that maintaining bots and web apps are the same effort-wise. An IRC can very easily be served of a framework that is event driven in nature -- on the lines of a node.js or eventmachine with barely any additional effort. On the other hand, doing this for a web application would require extensive effort for writing your code the right way and bringing in the right tools like a templating engine. You will build your own framework almost.

2) Since the application is process heavy and take over 45 seconds to serve a request, this would involve extensive tweaking of your forward proxy which in the case of most data centres and clouds will simply respond with a "Service Unavailable" after 30 seconds. Off course you can defer this to a background job and provide the results elsewhere on your application. But like I said, you would be doing things that the user actually does not need.

3) How would you scale up such an operation heavy application if it were a web app. Have 1000 background jobs running for a 1000 users? An event driven approach will pay off for sure. And 1) illustrates why its easier to do this with a bot.

The real disadvantage I came across was to be able to get it to work with a load balancer. I can write about this if there is more interest.