We’ve achieved our goals already with this series: we have a web service written in Rust, built into a Docker image with nix, with a nice dev shell, that we can deploy to fly.io.

But there’s always room for improvement, and so I wanted to talk about a few things we didn’t bother doing in the previous chapters.

Making clash-geoip available in the dev shell

There it is. The final installment.

Over the course of this series, we’ve built a very useful Rust web service that shows us colored ASCII art cats, and we’ve packaged it with docker, and deployed it to https://fly.io.

We did all that without using nix at all, and then in the last few chapters, we’ve learned to use nix, and now it’s time to tell docker build goodbye, along with this whole-ass Dockerfile:

In the previous chapter, we’ve made a nix “dev shell” that contained the fly.io command-line utility, “flyctl”.

That said, that’s not how I want us to define a dev shell.

Our current solution has issues. I don’t like that it has import <nixpkgs>. Which version of nixpkgs is that? The one you’re on? Who knows what that is.

Also, we haven’t really seen a mechanism to use .nix files from elsewhere.

Remember the snapshot we made allll the way back in Part 1? Now’s the time to use it.

Well, make sure you’ve committed and pushed all your changes, but when you’re ready, let’s go back in time to before we installed anything catscii-specific in our VM.

This should emulate the experience of a colleague onboarding onto the project well enough!

(I didn’t actually use VirtualBox’s snapshot feature for this, I actually set up a Ubuntu 22.10 VM on another computer entirely, but the effect should be much the same).

Now that my website is deployed as a container image, I wanted to give nix a try. I’m still doing it the old-fashioned way right now: with a Dockerfile, running cargo in a “builder” image, copying stuff out of there into a slimmer image (that still has an Ubuntu base, even though distroless images are a thing now).

But why?

I was mostly interested in nix because some parts of my website have pretty big native dependencies. futile itself mostly relies on sqlite3 and some JS engine (used to be quickjs, currently duktape because MSVC Windows builds). But the asset processing pipeline, salvage (which I’d like to integrate with futile at some point) has a bunch more!