Robert Postill

I wanted a badge on Github so visitors to the repo would know immediately that they were dealing with something tested. Racket's rackunit testing library had a pre-existing integration for coveralls and it was quick and easy to get it going.

I don't use much of the fancier parts of coveralls but it does exactly what it says on the tin. Tracks your coverage, reports it faithfully over time.

I wanted identity to be done well. So I knew I needed an identity provider. In order to make that decision there were the obvious cloud provider offerings and independent offerings.

I've used Auth0 about five different times now and I'm really comfortable with it. I've heard grumbles about Cognito being limited in functionality. Also having been around Azure Active Directory I can tell you its complexity is a significant barrier to entry. Certainly if you're using anything that isn't the dotnet framework.

Auth0 uses JWT and OAuth2 under the hood, so if you need to you can rely on tools for those technologies to integrate. Which is great. The docs are generally thorough and there are examples you can use to get going. Finally, the experience I have means it is quick for me to get going and make the right configuration choices.

I chose Make in part because the Racket community has make files for projects that were easy to use and adapt to my project. The reality is Make is still the best cross-platform cross-ecosystem build tool. Sure you could use something more tailored to your language or framework. But who has a stack so simple it only has one language or framework in it? Not me, certainly.

Basically Docker provided us with more deployment options and less security issues.

The most important part of Docker is repeatable builds. Once you have repeatable builds you gain the ability to secure the image. That comes from stamping the image, providing SBOMs and being able to assure clients the thing we said we shipped is the thing running in their environment.

Because we needed to manage things that weren't in Azure we went with Terraform.