class thread_queue::event
Declaration
class thread_queue::event : public async_event_impl { /* full declaration omitted */ };Description
Abstract base class for async_event implementations.
Inherits from: async_event_impl
Member Variables
- std::variant<std::future<completion>, completion> m_state
Member Function Overview
- event(std::future<completion> future) → explicit
- get_native_execution_time() → std::optional<std::chrono::nanoseconds>
- get_result() → void *
- is_complete() → bool
- get_completed() → completion *
Inherited from async_event_impl:
Member Functions
¶explicit event(std::future<completion> future)
explicit event(std::future<completion> future)Parameters
- std::future<completion> future
¶std::optional<std::chrono::nanoseconds>
get_native_execution_time()
std::optional<std::chrono::nanoseconds>
get_native_execution_time()Description
Returns the time execution time as measured if profiling was enabled in the issuing component. Requires is_complete() to be true.
¶void* get_result()
void* get_result()Description
There is only one instruction type which returns a result, namely alloc_instruction returning a pointer to the allocated memory, i.e. a void*. Having a void* return type on async_event_impl is somewhat leaky, but we don't gain much by wrapping it in a std::any.
¶bool is_complete()
bool is_complete()Description
If this function returns true once, the implementation must guarantee that it will always do so in the future. The event is expected to be cheap to poll repeatedly, and the operation must proceed in the background even while not being polled.
¶completion* get_completed()
completion* get_completed()