Delivering a simple payload to IRC is an ideal use-case for Function-based cloud computing. Last year when GitHub discontinued their service for pushing notifications into IRC channels, I had a perfect situation to couch-hack with a library I had recently discovered: Azure Functions for Rust

Functions-as-a-Service are a technology which I constantly kick myself for liking as much as I do. Functions, Lambdas, or whatever you wish to call them are a novel reinvention of FastCGI, and lots of fun to mock, but like its predecessors: when appropriately used, Functions can be incredibly useful. I personally think Functions work well as low-throughput “glue”. Binding simple services or components together which don’t warrant an “always-on” application means Functions can make good technical and financial sense.

I wrote this simple function a few months ago, before I actually knew what I was doing with Rust. In the span of an evening I was able to cobble something together which would read incoming JSON and write information into a Freenode channel:

pub fn webhook(req: HttpRequest) -> HttpResponse {
    if let Ok(event) = req.body().as_json::<PushEvent>() {
        info!("Parsed push event from {}", event.sender.login);
        return "Messages dispatched to IRC".into();
    else {
        return "Failed to parse event".into();

(it’s really simple, I swear)

I ended up learning more during this exercise about the Azure Functions Runtime For Linux, than Rust itself. Traditional Azure Functions run inside a Windows context, which aside from being totally hilarious, is incredibly limiting for people wanting to do anything serious that isn’t built around C#. The Azure Functions for Linux support is exactly how I wish all Functions-as-a-Service providers operated: just build the thing into a Docker container. While the Dockerfile is a bit esoteric, the end result is a conventional container which can be shipped off to a registry from a CI process, just like practically everything else in my world.

Overall the “irc-push-function” I developed is simple enough, runs without a hitch, and hasn’t cost much of anything the entire time it has been running. If you’re already running workloads in Azure, I certainly recommend trying out azure-functions-rs.