I built the prototype for Dead Man’s Switch over several off-and-on months using a smorgasbord of technologies that I wanted to learn or improve in.
The project started with a React front-end. As I was already reasonably familiar with React, I decided to spice it up by using Tailwind (which I hadn’t ever used before) as my CSS framework. I liked using Tailwind a lot, but as someone who doesn’t like to make a ton of decisions in the front-end, I found it at times to be somewhat tedious - especially when compared to something like Bootstrap (which I’ve used a lot of in the past).
For the back-end I decided to use NestJS - it was another project that I hadn’t used before but had heard good things about. I was (and still am) interested in moving towards a more service-oriented architecture after working exclusively in monoliths and Nest seemed to be the most feature-rich Node framework that was built with services in mind. I really enjoyed using Nest and despite some idiosyncracies (namely how the usage of decorators) - I think I’ll continue using it for projects in the future.
I decided to use GraphQL instead of REST in part because Nest supported it so well - but also because it was a service that I had never set up from the provider side. It was definitely overkill for this simple project - and there’s just some simplicity to REST that I love - but I do intend on continuing to use it in the future when I can.
The one design decision I made for simplicity’s sake was to use a MongoDB.
I initially was building this project with separate front-end and back-end repositories, but at a certain point that became tedious, so after doing some reading about how to best handle this problem, I was pleasantly surprised that Nest had built-in Monorepo capabilities. While the way that I brought in my front-end code to my codebase wasn’t perfect, Nest made it really easy and handled a lot of the lifting for me.
Finally, I wanted to use this project to extend some of my dev-ops knowledge, so I used Docker to containerize the project and ultimately used Kubernetes (with Helm’s help) to orchestrate. I ended up using Okteto for PR builds (and because they have a free tier) and am super happy with their service.
As for the future, I’m not sure what if anything I’ll end up doing with this project. I intended to get it to a production environment and use it for at least personal use (and allow folks to store small amounts of data), but more I thought about it, the more terrified of maintaining it I became. This is a project that (at least theoretically) lasts for lifetimes. And it’s a project that would store potentially incredibly important data.
While perhaps some day I may have the capacity to commit the rest of my life to maintaining a project, that day is not today, so for now this project will live as it is.
Thank you for making it this far. You can view the project here.