Conference talks revisited

One of the most important things is being willing to admit when you're wrong.

Over the years, I've given lots of conference talks and with them I've given lots of advice.

Looking back, a lot of it has turned out to be bad advice.

Here I wanted to take a second to look through some of these talks and mention how my thinking on them has changed.

Better JSON through streams

In this talk, I looked at a library called Oboe.js for loading and parsing JSON. In it I recommended using this library so that you can access data from a big JSON object before all of it has been parsed and ready to use in memory.

In hindsight, I don't think there are many scenarios where you would want to do this.

On the browser side, Oboe replaces whatever other method you would use for fetching data and allows you to write a DSL to subscribe to events when certain nodes in that data are found.

If you find yourself sending so much data over the network that it can't be easily parsed out by your client, it is probably a better solution to evaluate how you're sending that data. If the data isn't all available at once and you would like to make it available in an event-driven interface, you might be better served by reworking your server-side to expose a websocket.

On the server side, the case is pretty similar. If you find yourself needing to represent data in a stream, there are many good tools for doing that and you probably don't want to add an extra DSL to do that.

The concepts behind streaming data transfer are very good to understand and this talk served as away for me to learn about them. If I were to give this talk again, I would drop Oboe.js completely and instead focus on event architectures and how streaming is just an implementation of that.

Elm => Javascript

Test Driven CSS

Pattern Matching in Javascript

Enter the Src, VR on the web

Where have I mentioned this?