Overview
I’ve been loosely planning a series of screencasts for a Nerves based project for about a year and a half now. I believe i finally have enough content and time to start it.
This is going to be a full stack, open source “Internet of Things” project. Here is an overview of all the parts:
A wireless sensor system
This will be the main feature we build out. Many sensors and actuators will be able to report back
a central “hub” device (our Nerves application). Plans so far include:
- Temperature
- Humidity
- Light level
- Remote relays
With the capability of more being added easily. Each sensor device will be able to receive remote
firmware upgrades, but otherwise be fairly “dumb”. These devices will not themselves be Nerves
devices, instead they will be Arduino based devices to keep costs down and encourage battery life.
A central “hub” or “gateway”
This will be the Nerves application. The sensors and actuators will report their data and receive commands
to and from this “hub” device. Initially, we will target the Raspberry Pi 0 W, again to keep costs down. Eventually we may introduce support for other supported devices.
The device will be developed “offline first”. This means it will be “connected”, but will
still work when disconnected. The fleet of devices will also be managed by NervesHub. This will be how we manage pushing updates to the deployed devices. (IE devices that would have been purchased)
A cloud web application
This will be a Phoenix based web app. I personally am not a web or JS developer, so it will likely not look
the greatest at first. The backend services will be the main focus of the series. Things the web app should
be in charge of:
- User registration/auth
- Device registration/auth
- Serving/receiving RESTful assets
- Serving a frontend of some sort
- Handling many simultaneous device connections via Phoenix channels
- Handling many simultaneous user/frontend connections via Phoenix channels
- ???
Initially the app will likely be deployed on DigitalOcean or Vultr. We won’t use Heroku or Gigalixir because
they terminate SSL connections before they reach the application layer. We will need SSL to validate devices against NervesHub. This will greatly simplify the logic required by the web app to validate devices.
Target Audience
No Nerves knowledge should be required, but I will likely make some assumptions when getting started with the streams. I’ll assume a basic knowledge of Elixir at a basic level. We won’t be doing anything “advanced” in either main applications, but a basic knowledge of Elixir syntax and how to work within Mix and ExUnit will be expected. I’ll also assume a basic knowledge of Phoenix or at least how an MVC framework works. I myself am certainly not an expert in Web Apps or Phoenix, so this is definitely not required.
Name
I haven’t started anything on this yet. This includes a name, domain, etc. I plan on doing all the bootstraping live. That said i could use some help picking a name for this project.
Timeline
There will probably be more casts at first to get the app bootstraped and to handle “devops” things like setting up the domains, Droplets, etc. These first few streams will likely be a little dense. Once up and running, I will likely do one or two screencasts a week, focusing on one or two small features at a time. This is mostly to prevent any burnout, and depending on how i feel, we may do more or less than this.
I will be streaming live on Twitch as I’ve used it before (I’m open to other streaming sites if someone makes a good argument). I will also publish recorded casts to YouTube as I don’t believe Twitch stores any streams once offline. I haven’t decided when the first stream will be, but I’m hoping to get started some time this week, or next. I’m thinking i will stream around 6PM Pacific time, but am open to suggestions for what time works best for the majority of users.
Licensing and other Copyright stuff
I plan on releasing all code as either MIT or APACHE-2. A final decision will be made upon first stream.