Some, such as atomic compare-and-swap, are both read-like and write-like.
</p>

<p>
A <i>goroutine execution</i> is modeled as a set of memory operations executed by a single goroutine.
</p>

<p>
<b>Requirement 1</b>:
The memory operations in each goroutine must correspond to a correct sequential execution of that goroutine,
given the values read from and written to memory.

	// non-profitable and allows to confirm that spinning does not do harm.
	// To achieve this we create excess of goroutines most of which do local work.
	// These goroutines yield during local work, so that switching from
	// a blocked goroutine to other goroutines is profitable.
	// As a matter of fact, this benchmark still triggers some spinning in the mutex.
	m := NewDRWMutex(ds, "")
	var acc0, acc1 uint64

			return buf.Bytes(), err
		}
	case madmin.ProfilerGoroutines:
		prof.ext = "txt"
		prof.record("goroutine", 1, "before")
		prof.record("goroutine", 2, "before,debug=2")
		prof.stopFn = func() ([]byte, error) {
			var buf bytes.Buffer
			err := pprof.Lookup("goroutine").WriteTo(&buf, 1)
			return buf.Bytes(), err
		}
	case madmin.ProfilerTrace:
		dirPath, err := os.MkdirTemp("", "profile")

	for i := 0; i < numBuckets; i++ {
		res.buckets[i] = bucket[K, V]{entries: make(map[K]*Entry[K, V])}
	}

	// enable deleteExpired() running in separate goroutine for cache with non-zero TTL
	//
	// Important: done channel is never closed, so deleteExpired() goroutine will never exit,
	// it's decided to add functionality to close it in the version later than v2.
	if res.ttl != noEvictionTTL {
		go func(done <-chan struct{}) {

// panicking goroutine. Executing a call to recover inside a deferred
// function (but not any function called by it) stops the panicking sequence
// by restoring normal execution and retrieves the error value passed to the
// call of panic. If recover is called outside the deferred function it will
// not stop a panicking sequence. In this case, or when the goroutine is not
// panicking, recover returns nil.
//

## Coroutine'ler { #coroutines }

**Coroutine**, bir `async def` fonksiyonunun döndürdüğü şeye verilen süslü isimdir. Python bunun bir fonksiyona benzer bir şey olduğunu, bir noktada başlayıp biteceğini bilir; ama içinde bir `await` olduğunda dahili olarak duraklatılabileceğini ⏸ de bilir.

// returns the slice of errors from all function calls.
func (g *NotificationGroup) Wait() []NotificationPeerErr {
	g.workers.Wait()
	return g.errs
}

// Go calls the given function in a new goroutine.
//
// The first call to return a non-nil error will be
// collected in errs slice and returned by Wait().
func (g *NotificationGroup) Go(ctx context.Context, f func() error, index int, addr xnet.Host) {

	}

	// Load exist schema cache, return if exists
	if v, ok := cacheStore.Load(schemaCacheKey); ok {
		s := v.(*Schema)
		// Wait for the initialization of other goroutines to complete
		<-s.initialized
		return s, s.err
	}

	var tableName string
	modelValue := reflect.New(modelType)
	if specialTableName != "" {
		tableName = specialTableName

			bucketCh <- b
		}
	}
	xioutil.SafeClose(bucketCh)

	bucketResults := make(chan dataUsageEntryInfo, len(disks))

	// Start async collector/saver.
	// This goroutine owns the cache.
	var saverWg sync.WaitGroup
	saverWg.Add(1)
	go func() {
		// Add jitter to the update time so multiple sets don't sync up.

Часто всю функциональность использования асинхронного кода с `async` и `await` кратко называют «сопрограммами». Это сопоставимо с ключевой особенностью Go — «goroutines».

## Заключение { #conclusion }

Вернёмся к той же фразе:

> Современные версии Python поддерживают **«асинхронный код»** с помощью **«сопрограмм»** (coroutines) и синтаксиса **`async` и `await`**.

Теперь это должно звучать понятнее. ✨