Thanks everyone.
I thought about the following design:

Inside the thread itself, I call m_promise.set_value() after the task finishes:

std::thread([this]()

{

task();

m_promise.set_value();

});

Then, when I want to stop the thread, I do:

if(m_thread.joinable()) {

if (m_future.valid())

m_future.wait(); // Wait until the thread signals it has finished

m_thread.join();

}

This way, I guarantee that by the time I call join(), the thread has already completed its task.

Would there be any issue with this implementation?

Also, if I want to make it even safer, I thought about adding:

if(!m_alreadyJoining.exchange(true))

m_thread.join();

to protect against joining twice in case of concurrent interrupts.

any feedback ? or what can be an issue here?