Skip to content

NoAvailableAlias/nano-signal-slot

Repository files navigation

nano-signal-slot

Pure C++17 Signals and Slots

Include

// #include "nano_function.hpp"         // Nano::Function, Nano::Delegate_Key
// #include "nano_mutex.hpp"            // Nano::Spin_Mutex, all threading policies
// #include "nano_observer.hpp"         // Nano::Observer
#include "nano_signal_slot.hpp"         // Nano::Signal

Declare

// Declare Nano::Signals using function signature syntax
Nano::Signal<bool(const char*)> signal_one;
Nano::Signal<bool(const char*, std::size_t)> signal_two;

Connect

// Connect member functions to Nano::Signals
signal_one.connect<&Foo::slot_member_one>(foo);
signal_two.connect<&Foo::slot_member_two>(foo);

// Connect overloaded member functions (required template syntax)
signal_one.connect<Foo, &Foo::slot_overloaded_member>(foo);
signal_two.connect<Foo, &Foo::slot_overloaded_member>(foo);

// Connect a static member function
signal_one.connect<&Foo::slot_static_member_one>();

// Connect a free function
signal_two.connect<&slot_free_function_one>();

Fire / Fire Accumulate

Slot emission is not guaranteed to be in the order of connection.

// Fire Signals
signal_one.fire("we get signal");
signal_two.fire("main screen turn on", __LINE__);

std::vector<bool> statuses;
auto accumulator = [&](bool srv)
{
    statuses.push_back(srv);
};

// Fire Signals and accumulate SRVs (signal return values)
signal_one.fire_accumulate(accumulator, "how are you gentlemen");

Disconnect

// Disconnect member functions from Nano::Signals
signal_one.disconnect<&Foo::slot_member_one>(foo);
signal_two.disconnect<&Foo::slot_member_two>(foo);

// Disconnect overloaded member functions (required template syntax)
signal_one.disconnect<Foo, &Foo::slot_overloaded_member>(foo);

// Disconnect a static member function
signal_one.disconnect<&Foo::slot_static_member_one>();

// Disconnect a free function
signal_two.disconnect<&slot_free_function_one>();

// Disconnect all slots
signal_two.disconnect_all();

Connection Management

Automatic connection management requires public inheritance from Nano::Observer<>.

struct Foo : public Nano::Observer<>
{
    bool slot_member_one(const char* sl) const
    {
        std::cout << sl << std::endl;
        return true;
    }
	...

Function Objects

Connected function objects must live longer than the connected signal.
(be sure to disconnect the function object prior to it destructing)

auto fo = [&](const char* sl)
{
    std::cout << sl << std::endl;
    return true;
};

...

// Connect any object that defines a suitable operator()
signal_one.connect(fo);

...

// Disconnect that same functor instance
signal_one.disconnect(fo);

Threading Policies

Nano-signal-slot has the following threading policies available for use:

  ST_Policy TS_Policy ST_Policy_Safe TS_Policy_Safe
Single threading only X - X -
Thread safe using mutex - X - X
Reentrant safe* - - X X

* Reentrant safety achieved using emission copying and reference counting.

Threading Policies - Aliases

When integrating nano-signal-slot, it is recommended to alias the Nano::Signal and Nano::Observer template classes.
When using a non-default Policy you must make sure that both Signal and Observer use the same policy.

namespace Your_Namespace
{

// Creating aliases when using nano-signal-slot will increase the maintainability of your code
// especially if you are choosing to use the alternative policies.
using NanoPolicy = Nano::TS_Policy_Safe<>;

template <typename Signature>
using Signal = Nano::Signal<Signature, NanoPolicy>;

using Observer = Nano::Observer<NanoPolicy>;

}

// Then use it via your namespace
Your_Namespace::Signal<bool(const char*)> signal_one;

Threading Policies - Mutex Policy

Both the TS_Policy and TS_Policy_Safe allow for a provided mutex type if non-trivial locking is desired.
The provided mutex type only needs to implement the BasicLockable requirement.

using Your_Policy = Nano::TS_Policy<Your_Mutex>;

Deadlock Disclaimer

The TS_Policy does not mitigate any deadlocks that could occur due to slot emissions fiddling with their signals. Additionally, when using this policy, it is not safe to destruct connected Nano::Observers from different threads. Generally if the threading posture in your application allows for it then use the TS_Policy for the performance. If the lifetimes of Signals vs Observers is unknown or if the slots could be hostile then use the TS_Policy_Safe.

Links

Benchmark Results Benchmark Algorithms Unit Tests