fix(doctor): probe podman too — a Podman machine was reported as "docker is not installed"

[MendixMau]

Direct lane (bin/)

What changed and why (one paragraph):

bin/doctor.sh advertised Podman in its advice text — "Rancher Desktop or Podman (docker-CLI compatible) and colima (macOS) are common substitutes" — while every probe in the section ran docker only: docker_daemon_up() ran docker info, the probe block gated on command -v docker, the not-installed branch warned "docker is not installed", and docker_start_hint() suggested open -a Docker. So a machine running Podman without the docker shim is fully able to run the container lane (mxcli docker check, project-bin/test-stack-up.sh) and doctor reported it had nothing — and on a team that cannot licence Docker Desktop, that reads as "go install software you are not allowed to have". This PR makes detection and reporting runtime-aware: prefer docker when present (including when it is a Podman shim — transparent and correct), fall back to podman, and name the runtime that answered in the output (ok podman responding — …). The bounded background/poll/kill probe and its exit contract (0 up · 1 down · 2 no answer within the bound) are preserved unchanged and now cover Podman too — a stopped podman machine stalls exactly like a stopped Docker Desktop, which is what the bound exists for. container_start_hint() suggests podman machine start (macOS/Windows) or systemctl --user start podman.socket (Linux/WSL) for the Podman case, and the not-installed warning names both runtimes with Podman as the licence-free option instead of implying Docker Desktop is required. MXTK_CONTAINER_RUNTIME=<name> forces one. Scope is detection and reporting only — how the gate invokes the runtime (mxcli docker check) is mxcli's own code in a different repo and is untouched. --install still auto-launches Docker Desktop on macOS only; Podman is deliberately left alone, because podman machine start on a machine that has never run podman machine init is not a safe guess to make on someone's behalf. Same defect class as #128: an instrument reporting a problem that does not exist.

Field evidence — which real project this ran against / was observed in, and what it measured or fixed there:

No field run. Be clear about this: the authoring container has no working container runtime, so this was NOT executed against real Docker or Podman. Every branch was exercised against fake docker/podman binaries on PATH — daemon up, daemon down, hanging daemon (bound bites), no runtime at all, and podman-only with docker absent from PATH — which proves the control flow and the strings, not the runtimes. The run that would prove it: on a Mac with Podman installed and no docker on PATH, bin/doctor.sh should report podman responding (or a podman machine start hint), where today it reports "docker is not installed". The field data point that motivated it is second-hand and named in the code comment: a colleague's "machine ready" status listed "Docker not installed" as a known issue; they may well have had Podman all along. Driver: the Mendix migration team cannot use Docker Desktop for licensing reasons and is standardising on Podman.

• No client data anywhere in the diff
• Size cap: 3 files, ~99 insertions / ~50 deletions (1 line of it CHANGELOG.md)
• Test tier reached: T0/T1 — bash -n bin/doctor.sh clean; bin/check-portability.sh clean (163 files inspected, no new violations); bash 3.2-compatible (no namerefs, no declare -A, no mapfile). tests/wave2/test-doctor-docker-probe.sh was not run (CLAUDE.md: scoped testing, ask first) — instead its four scenarios were replicated by hand against the changed script, and every one of its grep strings reproduces verbatim: probing the docker daemon (bounded: 2 s, gave no answer within 2 s, MXTK_DOCKER_PROBE_SECS=30, Then re-run: bin/doctor.sh, daemon is not responding, To start it:, docker daemon responding, docker probe skipped, and docker daemon responding correctly absent from the hang case. The fixture only ever puts a fake docker on PATH, so detection resolves to docker and the label is docker daemon — its assumptions are intact. A maintainer run of bash tests/wave2/test-doctor-docker-probe.sh bin/doctor.sh would confirm that mechanically; CI runs it anyway.
• For a changed instrument: both layouts irrelevant here (no path resolution touched); both platforms considered — the Podman start hint branches on PLATFORM (macos/gitbash → podman machine start; linux/wsl → podman.socket), and no new external tool is invoked beyond <runtime> info
• CHANGELOG.md line appended in this PR, crediting the team's Docker Desktop licensing constraint
• No new bug entries or skills in this PR

Also touched: project-bin/_common.sh — one comment line, the stale cross-reference to bin/doctor.sh's docker_daemon_up (now container_daemon_up). No behaviour.

Noticed, deliberately not changed: the --no-docker flag, the MXTK_DOCTOR_SKIP_DOCKER / MXTK_DOCKER_PROBE_SECS env names and the "docker probe skipped" message keep their docker-shaped spelling — renaming them is a compatibility break for existing callers and fixture assertions, for no reporting benefit. mxcli docker check and project-bin/test-stack-up.sh still assume a docker CLI; whether they work under bare Podman is a separate question in a separate repo, and this PR does not claim they do.

🤖 Generated with Claude Code

https://claude.ai/code/session_0187r9bQWkZAEppyEqrDevjK

---

Generated by Claude Code