setup_check
diapason.desktop.setup_check
¶
What dictation needs to work, as inspectable data.
Every failure we hit getting dictation running looked identical from the outside — "nothing happens" — and each had a different cause: a missing permission, a missing model, an agent that was not loaded. The checks are therefore individually reportable: the setup wizard, the doctor command and any future UI all read the same list, and each item says what is wrong AND what to do about it.
Pure enough to test: each check returns a verdict, and only the probes it calls touch the system.
Check
dataclass
¶
One requirement, its state, and the single next action.
check_hotkey(config=None)
¶
Is the configured hotkey one the tap can actually bind?
check_model(config=None)
¶
Is a speech backend actually resolvable right now?
check_permissions()
¶
The three TCC grants, reported separately.
Separately, because they fail separately and each has its own pane — lumping them into one "permissions OK" line is what made the microphone failure so hard to see.
check_service()
¶
Is the background agent installed and loaded?
is_ready(checks)
¶
Ready means: dictation would work right now if you held the key.
run_all(config=None)
¶
Every check, in the order a first-time user should fix them.