So, Today we are trying something different. LIVE JOURNAL.

A little hummor from a sandbox.

Session 1 — Google Assistance

First session was on Google Assistant Integrations and Actions. There’s some real interesting possibilities in this space right now. The gist of setting up the action was very straightforward.

First, set an app in the API(As you do with most things.)

Great a very simple API with some Oauth2 handles and functional callbacks.

Chain these functions together.

This was all very straightforward but they did bring up a very interesting point. When you are writing you responses assume that you are writing lines 1, 3, and 7 of an exchange. You need to make sure that the other speaker, in this case your customer, is able to hold the conversation. This should be done with only questions. I’m looking forward to playing with it.

Session 2 — IoT

Sadly work snagged me to do some support while I was in this talk. My main take away, other than I can’t wait to hack on my Android Things dev kit? Don’t use cheap solder and there are some things I need to learn about board fab.

I mean, I understand gerbers and what that means, but its not easy. Android Things takes care of some of this by making sure that your hardware is based upon a SoC that has already taken care of most of the testing that could hold up your hardware in testing but the steps to get there require a level of test proofing that I don’t think most software dev just don’t have.

This said, I sympathize with the burning of one’s fingers.

Session 3 — TENSOR FLOW

Okay, this the big one for today. Tensor Flow. Machine learning. THE FUTURE!

The first speaker was on how easy Tensor Flow was to get running. I’ve done this, you’ve done this, every one had done the Regressor tutorial on this. The new feature here was using the TensorFlow environment to spread your load without doing any new work on your part. This makes me think of easy load balancing(TM) that you get in a kubernetes cluster.

The second speaker, with an amazing french accent, was on the Keras modeling system. Presented as a ‘lego set’ for ML, it looks very powerful. Its a level of abstraction over the normal TensorFlow models that takes a lot of the best practice work out of the way. If I had to liken it to something else, it feels like a wrapper around low level file handling systems. In a good way. You wouldn’t write a direct read/write I/O buffer in Node, would you?

Don’t answer that.

Session 4 — LUNCH

Session 5 — Angular CLI PWA

So, lunch took a little longer than planned. But it looks like the CLI has most of the tools we’ll need. ng-pwa-tools looks awesome. Looking at the build tools almost all of their could be moved to a Dockerfile or even gulp.

Not only this, but we’ve got some totally amazing tools to caching built in now. These are a real game changer in that it removes the work of data caching on device.

Okay…. So yeah. That was…. intense.

To summarize, we’ve got a series of tools now to build PWAs that doesn’t require you to work too hard. Not only this, but you can prototype in Firebase and then move to your cluster. It uses HTTP/2 to send the chunks ahead of time. Not only this, but App Shell allows you had a loading screen based upon your normal page!


SESSION 6 — IoT with Android Studio

Last night I had a chance to play with Android Studio 3.0 Canary. For this talk they are going over the Android Things and how to hook up the various drivers. If you had done anything with GPIO and Node on Raspbian then its all old hat.

The big change that you could have noticed is that they are some nice drivers and easy threading. I’ve always had mixed opinions about threading on android/IoT. On a rPi Zero you could just write the whole thing in node, let node schedule it, and be done. The issue with this is that when you DO get things working with user input you’ve got some major issues with Node concurrency logic. Not all people can do that.

For the flip side, threading can be VERY powerful. And if you use Android Things then that means you get the benefits of Google’s optimizations passes. And power use/cycle optimization is a HUGE concern when it comes to IoT devices.

That all said, the speakers are GREAT.

Session 7 — Web Security


One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.