# Running and stopping a program

[Programming guides](README.md)

Ordinary top-level setup statements execute in source order while top-level
`when` statements register event scripts. After setup, the runtime runs the
registered `started` scripts, arms every one-shot `after` script, and processes
queued message work cooperatively. A program normally finishes when neither
ready nor sleeping scripts remain, unless persistent key or pointer handlers
keep it waiting for input.
`exitApp()` finishes the complete program immediately and successfully, even
from a nested function, loop, or event script. Remaining work does not run.

In an [IDE](../concepts/ide.md) or browser preview, `exitApp()` finishes only the learner program's
execution session and leaves the preview available for inspection. A native
graphical program that naturally reaches its final statement retains its last
frame until the user closes the window; `exitApp()` instead requests normal
host shutdown and resource cleanup.

`return` exits only the current function. IDE Stop is external cooperative
cancellation and is not Pliro syntax.

Positive `wait` calls and completed input requests are cooperative checkpoints.
They keep Stop and graphical event processing responsive and start a fresh
instruction-budget slice. In an event script, `wait` parks only the active
script. `broadcast` queues exactly matching message scripts and yields the
active script so the deterministic [FIFO](../concepts/event-loop.md) scheduler can continue other ready
work. Put a persistent animation loop in `when started():` when other event
handlers need to run during its waits; ordinary top-level setup must finish
before the event-scheduler phase begins.

```pliro
# language: en
let running = true

while running:
    if keyDown("space"):
        exitApp()

    # Update and draw one frame here.
    wait(0.04)
```

## Related entries

[Syntax and values](../syntax/README.md) · [Built-ins](../builtins/README.md) · [Programming guides](README.md)