I built boxr to stay rootless by default, with no always on daemon, and still feel familiar if you already use Docker or Podman commands. This article focuses on two everyday jobs: building an image from a Dockerfile, and running a small app with a Compose file. Both are where a new engine either earns trust or loses it.
What to try first: build a simple Dockerfile with boxr, then run it as your own user, with no sudo on the default path. Next, take a small Compose file with two services, bring it up, check the logs, and stop it. If you already run containers rootless, try the same files you use today and note where the behaviour differs.
What the project reports so far: the project test reports list the Compose suites as passing, and a cold Dockerfile build at about 332 milliseconds in the project Linux benchmarks. Those are project reported results, not a fresh independent rerun, and the next step is to rerun them at one pinned release, with repeats and variance, on a stated machine.
What is still slow, stated plainly: in those same project benchmarks, boxr was slower than Podman on cached builds, on single container startup, and on volume input and output. A merged startup latency improvement landed on main today, with a version bump to v0.1.71, but that change does not by itself close the gap, and it still needs a fresh benchmark at a pinned release before any new speed claim is made.
What to report back: which Dockerfile step behaved differently, which Compose feature worked or failed first, and what felt slow or unfamiliar. That feedback decides what gets fixed next, more than any headline result does.
Repository: https://github.com/kchaitanya863/boxr
