[Suggestion] 'restart --with-dependents' does not restart dependents if they are already running

[ericreeves]

I love seeing how wpmctl is shaping up.

As I've been relying on it for managing Komorebi, Yasb, whkd, Altsnap, and Masir, I've implemented the following 'desktop.toml':

[Unit]
Name = "desktop"
Description = "Everything I need to work on Windows"
Requires = [
    "yasb",
    "whkd",
    "komorebi",
    "altsnap",
    "masir",
]

[Service]
Kind = "Simple"
Autostart = true
Restart = "Never"

[Service.ExecStart]
Executable = "cmd.exe"
Arguments = ["/c", "pause"]

With this configuration, I would intuitively expect "wpmctl restart -d desktop" would fully stop/start all of the dependent services, but it does not. It WILL start them if they are stopped/failed, though.

2025-06-02T15:56:59.868268Z  INFO wpmd: received socket message: RestartWithDependents(["desktop"])
2025-06-02T15:56:59.868369Z  INFO wpmd: successfully queued socket message
2025-06-02T15:56:59.868442Z  INFO wpm::process_manager: desktop: stopping unit
2025-06-02T15:56:59.868594Z  INFO wpm::process_manager: desktop: sending kill signal to 42484
2025-06-02T15:56:59.870610Z  INFO wpm::process_manager: desktop: process 42484 successfully terminated
2025-06-02T15:56:59.870707Z  INFO wpm::process_manager: desktop: requires yasb
2025-06-02T15:56:59.870821Z  INFO wpm::process_manager: desktop: requires whkd
2025-06-02T15:56:59.870908Z  INFO wpm::process_manager: desktop: requires komorebi
2025-06-02T15:56:59.870985Z  INFO wpm::process_manager: desktop: requires altsnap
2025-06-02T15:56:59.871108Z  INFO wpm::process_manager: desktop: requires masir
2025-06-02T15:56:59.871177Z  INFO wpm::unit: desktop: starting unit
2025-06-02T15:56:59.873365Z  INFO wpm::unit: desktop: running pid 43408 liveness healthcheck (1s)
2025-06-02T15:57:00.928557Z  INFO wpm::unit: desktop: passed healthcheck

Example if a service is not running (masir):

2025-06-02T15:57:48.311490Z  INFO wpmd: received socket message: RestartWithDependents(["desktop"])
2025-06-02T15:57:48.311588Z  INFO wpmd: successfully queued socket message
2025-06-02T15:57:48.311654Z  INFO wpm::process_manager: desktop: stopping unit
2025-06-02T15:57:48.311722Z  INFO wpm::process_manager: desktop: sending kill signal to 43408
2025-06-02T15:57:48.313694Z  INFO wpm::process_manager: desktop: process 43408 successfully terminated
2025-06-02T15:57:48.313784Z  INFO wpm::process_manager: desktop: requires yasb
2025-06-02T15:57:48.313918Z  INFO wpm::process_manager: desktop: requires whkd
2025-06-02T15:57:48.313998Z  INFO wpm::process_manager: desktop: requires komorebi
2025-06-02T15:57:48.314070Z  INFO wpm::process_manager: desktop: requires altsnap
2025-06-02T15:57:48.314128Z  INFO wpm::process_manager: desktop: requires masir
2025-06-02T15:57:48.314190Z  INFO wpm::unit: masir: starting unit
2025-06-02T15:57:48.317518Z  INFO wpm::unit: masir: running pid 53036 liveness healthcheck (1s)
2025-06-02T15:57:49.365251Z  INFO wpm::unit: masir: passed healthcheck
2025-06-02T15:57:49.365356Z  INFO wpm::unit: desktop: starting unit
2025-06-02T15:57:49.368648Z  INFO wpm::unit: desktop: running pid 34664 liveness healthcheck (1s)
2025-06-02T15:57:50.418196Z  INFO wpm::unit: desktop: passed healthcheck

In a similar vein, "stop --with-dependents" feels like it would with go along with this command/pattern nicely, and behave in the same way.

Thoughts?

[LGUG2Z]

I think this makes sense, once we have stop --with-dependents we can just call it as part of restart --with-dependents