hl.device() has no getter — a plugin cannot read back a device's enabled state

[roymelgarv]

Summary

hl.device({ name = <dev>, enabled = <bool> }) sets an input device's enabled flag at runtime, but there is no counterpart to read it back. Every path I could find either always reports the same value regardless of state, or doesn't expose the field at all.

Why this matters

I maintain a small Omarchy plugin (https://github.com/roymelgarv/omarchy-keyboard-cleaner) that disables input devices via hl.device() so a keyboard can be safely wiped down. The plugin tracks lock/unlock state itself, but has no way to confirm against Hyprland that a disable actually landed, or that an unlock actually cleared it — it can only trust its own bookkeeping. That matters more than usual here: a bug in that bookkeeping (which I did hit) can silently strand a real keyboard with no way for the plugin, or a person debugging it, to detect the mismatch by asking Hyprland directly.

What I tried (Hyprland 0.56.2)

• hyprctl devices -j → the JSON does carry an enabled key per device, but it is null for every device unconditionally — including while a device is demonstrably disabled via hl.device() moments earlier. It never reflects real state.
• hl.get_config('device:<name>:enabled') and hl.get_config('device[<name>]:enabled') → both return nil. Ruled out a broken call: hl.get_config('input:kb_layout') on the same session correctly returns the configured layout, so the reader itself works — device:enabled specifically isn't exposed through it.
• hyprctl getoption 'device[<name>]:enabled' → no such option.
• Dumped the whole hl.* table: there are getters for monitors, windows, workspaces, layers (get_monitors, get_windows, get_workspaces, get_layers, ...), but device itself is a bare setter with nothing beside it.

Related but distinct

#5724 covers hyprctl getoption not supporting device:* for config-declared values, closed as a known hyprlang-migration tradeoff. This is narrower and Lua-specific: even a value set purely at runtime through hl.device() (never touching a config file at all) has no way to be read back through the Lua API either. A hl.get_device(name) (or an enabled field that's actually populated in hyprctl devices -j) would close both gaps for anyone driving devices from a plugin.

Ask

A read path for a device's enabled state — ideally hyprctl devices -j actually populating the field, since that's the zero-migration fix for every existing consumer, but a hl.get_device() Lua getter would also work for plugins.

Happy to test against a branch if useful.

[jbouze]

Seconded, and same thing for mice please.

This would work:
https://github.com/jbouze/Hyprland/tree/showDevicesEnabled

Happy to send a PR if this goes in the direction the dev team wants to go.

[vaxerski]

#15903