Why exporting WHEELS_ENV=production doesn't flip your app into production — and everything else about the config pipeline: the three-file load order, the .env chain that shadows OS variables, what production mode really changes, and the reload gate's fail-closed mechanics.
Three small pieces of plumbing turn 'the site gave me an error around lunchtime' into a single grep: the RequestId middleware, the /up health endpoint your app already ships, and a forty-line JSON logging pattern. Plus the two health-endpoint mistakes everyone makes once.
The manual install that took the community two days to reverse-engineer in Wheels 3 is now two zip files and zero external dependencies. How to install, migrate, test, seed, and upgrade a Wheels 4 app with no CLI, no CommandBox, and no package manager — plus the new guides that document every step.
Wheels 4.0 ships with a new command: wheels deploy. It's a port of Basecamp's Kamal into the Wheels CLI — zero-downtime Dockerized deploys to Linux servers over plain SSH, no Ruby runtime required. This post covers what got ported, the one deliberate divergence from Kamal, and the byte-compatibility contract that lets you take over a Kamal-managed server without cleanup.
When a developer opens a pull request against the Wheels framework, it kicks off one of the most comprehensive CI/CD pipelines you'll find in any open-source project. Whether you're a first-time co...