- Validate PTY_TOOL at startup and verify the selected binary exists
- Add abduco_session_alive() using `abduco -l` instead of checking BRIDGE_SOCK
(abduco does not create a socket file, so the old check always failed)
- Fix wait_for_session, manage blocking loop, and attach to use tool-specific
liveness checks rather than grouping dtach and abduco together
- Add default (*) error branches to all PTY_TOOL case statements
- Check kill -0 BRIDGE_PID inside the port readiness loop to fail fast if the
bridge exits before its ports are ready
- Add SIGTERM/SIGINT trap with a cleanup() function in run mode so Docker stop
properly reaps bridge and socat children instead of waiting for the kill timeout
- Replace \$(hostname) with <container> placeholder in user-facing error messages
- Use golang:1.26-trixie builder instead of debian:sid
- Build proton-bridge from source via version argument/envelopment
- Add support for PTY tools (dtach, abduco, reptyr) for interactive sessions
- Introduce manage and attach commands for bridge CLI sessions
- Improve daemon startup with port readiness checks
- Add HEALTHCHECK and configurable CMD/ENTRYPOINT
- Harden entrypoint with strict bash flags and better error handling
- Install additional runtime deps (libfido2, procps) and optional PTY tools
Ran into this issue when trying to use in kubernetes via k8s-at-home/protonmail-bridge. Took a bit of figuring out before I realized I had to manually kill the other instance. The only output from the failed load is a dbus issue, and an immediate exit.
