Notes

Occasional writing about tooling and small experiments.

Keeping a build reproducible

Pinning versions is the easy part. The harder part is noticing when the environment drifts underneath a build that still reports success.

On measuring the wrong thing

A metric that is easy to collect tends to replace the one that mattered, and the substitution is rarely noticed at the time it happens.

Notes on backups

An untested backup is a guess. Restoring one on a schedule turns it into something you can rely on.