The client

The nvda fixture wraps the session’s NvdaClient in the DSL. Request nvda_session to use the client directly. Both expose the same namespaces: speech, braille, keys, config, log, and addons.

Member Behaviour

Member	Behaviour
`version`	`NvdaVersion` with `version`, `api_version`, `api_compat_to`, and `channel`
`wait_until_idle(*, timeout=10.0)`	blocks until NVDA’s event queue drains
`reset()`	clears speech, braille, and log, and restores the baseline configuration
`restart_harness(*, timeout=60.0)`	kills and relaunches the NVDA process; does not exercise NVDA’s own restart
`restart_nvda(*, timeout=60.0)`	triggers NVDA’s own `core.restart()` and reattaches; requires `allow-eval`
`eval(source)`, `exec(source)`	run code inside NVDA; require `allow-eval`
`exec_nowait(source, *, label=…)`	queues code on NVDA’s main thread without waiting; use it for scenarios that open a modal dialog
`simulate_modal(gesture="enter", *, timeout=10.0)`	closes a real modal dialog opened by `exec_nowait()`; returns `False` if none appeared

version

NvdaVersion with version, api_version, api_compat_to, and channel

wait_until_idle(*, timeout=10.0)

blocks until NVDA’s event queue drains

reset()

clears speech, braille, and log, and restores the baseline configuration

restart_harness(*, timeout=60.0)

kills and relaunches the NVDA process; does not exercise NVDA’s own restart

restart_nvda(*, timeout=60.0)

triggers NVDA’s own core.restart() and reattaches; requires allow-eval

eval(source), exec(source)

run code inside NVDA; require allow-eval

exec_nowait(source, *, label=…)

queues code on NVDA’s main thread without waiting; use it for scenarios that open a modal dialog

simulate_modal(gesture="enter", *, timeout=10.0)

closes a real modal dialog opened by exec_nowait(); returns False if none appeared

Use restart_harness() for a clean process and restart_nvda() to verify behaviour that lives in NVDA’s real self-relaunch path. See Closing a real modal dialog for the modal workflow.