Add condition variable wrapper #17
Conversation
In line with `fine::Mutex` and `fine::SharedMutex`, this commit adds a wrapper, `fine::ConditionVariable` for `ErlNifCond` with an API similar if not identifical to `std::condition_variable`.
| [[deprecated( | ||
| "usage of `void fine::ConditionVariable::wait(std::unique_lock<Mutex> " | ||
| "&, Predicate)` is preferred")]] | ||
| void wait(std::unique_lock<Mutex> &lock) noexcept { | ||
| enif_cond_wait(m_handle.get(), *lock.mutex()); | ||
| } |
There was a problem hiding this comment.
Should we even expose this function?
If we remove it, I am afraid a user will use enif_cond_wait directly without understanding that enif_cond_wait can return earlier than expected.
There was a problem hiding this comment.
Deprecation is not really accurate for this case, so I would either skip deprecation or remove the function.
The same concern goes for std::condition_variable::wait, right? So if we want to maintain API compatibility, we should probably include it. That said, I don't have strong opinion, so pick whichever approach you consider better :)
|
|
||
| std::thread thread4([&] { | ||
| notified = true; | ||
| cond.notify_one(); |
There was a problem hiding this comment.
Theoretically thread4 could run first and then none of the other threads needs to wait, because the condition is already true.
We can make the test more robust, by having a num_threads_counter and each thread after obtaining the lock does num_threads_counter++. Then thread4 can wait for num_threads_counter == 4, using another condition variable. wdyt?
In line with
fine::Mutexandfine::SharedMutex, this commit adds a wrapper,fine::ConditionVariableforErlNifCondwith an API similar if not identifical tostd::condition_variable.