Goodbye Heroku free plan

08 January 2023

A few of my posts on here have been about coding on the cheap. Static sites, EC2 instances with everything installed on them, Github actions, free plans. Sometimes it’s to my detriment as I sacrifice a whole lot of my time for a few less dollars per month. Buuuuuut my side-projects haven’t been aimed at making money, so I don’t want them to suck up money. Maybe after a few more years on a developer salary my priorities will change and I’ll shoot for options that save time over cash. But I also enjoy the limits I set myself. They narrow down the infinite options we have for building and hosting sites.

These last few months have been a sad time for the web app cheapskates out there. We’re all slaves to capitalism — especially companies with “sales” in their name. And on 25 August 2022 Heroku (now a subsidiary of Salesforce) announced that they would be ending their free plan. In their announcement they advised that starting 28 November 2022 they would start shutting down free-plan dynos. The app I wrote to learn Laravel, Plantr, was hosted on a free Heroku dyno. I started revisiting the site throughout the end of 2022. It continued serving content into December, but a couple of weeks ago I hit the site and got an error message from Heroku.

Building that app over the last quarter of 2020 introduced me to a whole bunch of concepts; I learnt about ORMs, Alpine.JS (even got a CSS tricks article showing off my form validation), dependency injection with controllers, and Composer. Using a Rails-style model-view-controller framework back in 2020 has come in quite useful now that I have a job at a C#/ASP.NET shop (I haven’t blogged about the .NET ecosystem, however I might start out of pure frustration).

I didn’t have a whole lot of proof that I knew how to code before I uploaded Plantr - maybe a couple of JavaScript games and this website. Now I’ve had a job in web development for almost a year, and I’ve written a few CSS Tricks Articles there’s slightly less pressure for me to run side projects to prove I can code. I will still do side projects but they’ll be for fun now.

I understand that expecting for-profit companies to continue hosting server-rendering applications is a bit of a tall order — even in this world where the cost of compute increasingly approaches zero. In Adam Wiggins’ interview on the ChangeLog he talked about how there were a whole bunch of hobby apps on Heroku and it was difficult to tell which free apps were actually important to people. Servers need to grind away to intermittently fire them up, and it takes storage to keep them. If there’s a whole bunch of one-off apps ready to run at any moment, that costs Heroku cash, possibly with zero benefit.

At the same time, it’s a little sad that my side-project is no longer out in the wild to show off, even if it might be to laugh about how much (or how little) I’ve learnt. Maybe one day I’ll give it a spruce-up and a few more features and upload it somewhere else again. If so, it would be my third attempt at making it. I don’t know if that’d be the best use of my time, so for now it's going off to application heaven. Plantr served its purpose for me as a self-teaching tool. I don’t think I want to pay $60.00 a year to keep it running somewhere else.

The web is such a democratic platform and I hope that the death of free plans doesn’t lock too many people out. Coming from New Zealand I’m far more economically privileged than a lot of the world. It’s hackers poorer than me we should worry about. But I reckon it's going to be okay: whichever platform provides the free plans and creates the least friction is the one that will going to get adoption.

That's the platform that’s going to be sustainable.