Life update
After a short 1 year, 3 months, and 29 day process, my green card has been issued.
It’s been a whirlwind since. We packed up life in Christchurch, moved to the Chicago North Shore and bought a house near the Chicago Botanic Garden, haggled for a car, started preschool, re-learned how to speak American, and generally pretended to be grown-ups.
Project update
Things are settling now though, and I’ve started working again on the map project I wrote about earlier in the year. I’ve been thinking about what I’d like to learn while also trying to avoid cognitive overload from everything being new and/or unrelated to what I’m doing at my normal job. That led to these decisions (more detail below):
Platform & language
- API and services: C#
- Web: Next.js
- Infrastructure: Kubernetes, Azure
- IAC: Pulumi
CI/CD
- Independent dev and prod environments
- Deployment via GitHub Actions
- Merge to
dev-> deploy to dev cluster (only appropriate components, eg, web, API, IAC) - Merge to
main-> deploy to prod cluster, manual approval step
- Merge to
- High-availability: min 2 instances of each critical component
Detail
Azure, C#, Pulumi
My day job is mainly C# and Azure, with Azure being fairly new to me which is why it’s been thrown into the mix in place of AWS. So far I’ve successfully deployed to Azure using Pulumi, which is similar to Terraform but written in your programming language of choice, as opposed to YAML for Terraform.
I hesitate to invoke the “t-shaped developer” concept, but I would like to focus more on going deep on a language. Migrating the existing work on the API from Java to C# is an attempt at that, as opposed to working on simultaneous projects across 2-3 languages like I have in the past.
Kubernetes
Kubernetes is definitely overkill for this, but also fun. I have some K8s experience (I helped migrate on on-prem site to K8s at Trade Me) and interested in building out a distributed system from scratch. Despite microservices/K8s falling out of favor a little bit recently, I see see K8s frequently mentioned as being used in interesting projects or companies I read about.
A use case!
The plan was always to build a service to fetch FAA aeronatical charts and display them to users. The rest, however, was vague: “some form of basic flight planning component”.
I had the thought while practicing instrument approaches in Microsoft Flight Simulator 2020, which I do to stay current when I’m not able to fly. Prepar3d (the older-version) had a post-flight debrief component built-in which was perfect for reviewing accuracy on instrument approaches or tracking airways, but for some reason it wasn’t ported to the newer version. I thought it’d be fun to build a replacement.
Although some do exist (Tacview, LittleNavMap, Volanta) they don’t quite fit what I’m looking for. The best I’ve found so far is tracking the flight in Foreflight (the best real-world-flying app in the world) and using their debriefing tools, but requires a Foreflight subscription and an iPad/iPhone to run on, which is probably a bit spendy for someone just wanting to debrief flight sim flights.
Either way, the point isn’t really to be a world-changing product - it’s to build something interesting, learn something new, just for fun.
Goals for the week
- Get minikube running locally
- Create dev containers for web and API, and get them running locally