Let's say that before activating the environment, the `PATH` variable looked like this:

//// tab | Linux, macOS

```plaintext
/usr/bin:/bin:/usr/sbin:/sbin
```

That means that the system would look for programs in:

* `/usr/bin`
* `/bin`
* `/usr/sbin`
* `/sbin`

////

//// tab | Windows

```plaintext

 * iterators that don't support {@code remove()}. Otherwise, all optional operations are supported.
 * Null row keys, columns keys, and values are not supported.
 *
 * <p>Lookups by row key are often faster than lookups by column key, because the data is stored in
 * a {@code Map<R, Map<C, V>>}. A method call like {@code column(columnKey).get(rowKey)} still runs

 * iterators that don't support {@code remove()}. Otherwise, all optional operations are supported.
 * Null row keys, columns keys, and values are not supported.
 *
 * <p>Lookups by row key are often faster than lookups by column key, because the data is stored in
 * a {@code Map<R, Map<C, V>>}. A method call like {@code column(columnKey).get(rowKey)} still runs

      // Note that if all the futures on the list are done prior to completing this loop, the last
      // call to addListener() will callback to setOneValue(), transitively call our cleanup
      // listener, and set this.futures to null.
      // This is not actually a problem, since the foreach only needs this.futures to be non-null
      // at the beginning of the loop.
      int i = 0;

 * iterators that don't support {@code remove()}. Otherwise, all optional operations are supported.
 * Null row keys, columns keys, and values are not supported.
 *
 * <p>Lookups by row key are often faster than lookups by column key, because the data is stored in
 * a {@code Map<R, Map<C, V>>}. A method call like {@code column(columnKey).get(rowKey)} still runs

# Build execution model

At the highest level, Gradle's execution model is quite simple:

Below is the protocol in some more detail:

1. The client looks for a compatible idle daemon. If there isn't one, it starts a new daemon.
2. The client connects to the idle daemon and sends it a request to do some work. If the daemon is no longer running, the client starts again.

	CMPPD	X0, X1, 4
	CMPPD	foo+4(SB), X1, 4

// LTYPEX spec10	{ outcode(int($1), &$2); }
	PINSRD	$1, (AX), X0
	PINSRD	$2, foo+4(FP), X0

// Was bug: LOOP is a branch instruction.
	JCS	2(PC)
loop:
	LOOP	loop // LOOP

// Tests for TLS reference.
	MOVL    (TLS), AX
	MOVL    8(TLS), DX

// LTYPE0 nonnon	{ outcode(int($1), &$2); }
	RET

/// tip | Dica

Se você se deparar com um `RuntimeError: Task attached to a different loop` ao integrar funções assíncronas em seus testes (e.g. ao utilizar o [MotorClient do MongoDB](https://stackoverflow.com/questions/41584243/runtimeerror-task-attached-to-a-different-loop)) Lembre-se de instanciar objetos que precisam de um loop de eventos (*event loop*) apenas em funções assíncronas, e.g. um callback `@app.on_event("startup")`.

This code won't be executed in your app, we only need it to *document* how that *external API* should look like.

But, you already know how to easily create automatic documentation for an API with **FastAPI**.

So we are going to use that same knowledge to document how the *external API* should look like... by creating the *path operation(s)* that the external API should implement (the ones your API will call).

/// tip

                int headerLine = i + 1; // 1-based
                // Look ahead to find [source, ...] until next sample or end of example block
                String sourceLang = null;
                for (int j = i + 1; j < lines.size(); j++) {
                    String look = lines.get(j).trim();

                    if (look.equals("[.multi-language-sample]") || look.equals("====")) {