recycle is an implementation of a simple C++ resource pool.
Table of Contents:
The recycle
project contains two types of resource pools:
- The
recycle::shared_pool
is useful when managing expensive to construct objects. The life-time of the managed objects is controlled by usingstd::shared_ptr
. A custom deleter is used to reclaim objects in the pool when the last remainingstd::shared_ptr
owning the object is destroyed. - The
recycle::unique_pool
works the same way as therecycle::shared_pool
but instead usesstd::unique_ptr
for managing the resources. Still we need a custom deleter - withstd::unique_ptr
this has to be part of the type. So thestd::unique_ptr
returned byrecycle::unique_pool
is of typerecycle::unique_pool::pool_ptr
.
Besides the fact that recycle::shared_pool
manages std::shared_ptr
and
recycle::unique_pool
manages std::unique_ptr
the API should be the
same. So in the following you can replace shared
with unique
to
swap the behavior.
The library itself is header-only so essentially to use it you just have to clone the repository and setup the right include paths in the project where you would like to use it.
The library uses C++14 features, so you need a relatively recent compiler to use it.
There are two ways we can control how objects are allocated:
Example:
#include <recycle/shared_pool.hpp> #include <cassert> struct heavy_object { // ... some expensive resource }; recycle::shared_pool<heavy_object> pool; // Initially the pool is empty assert(pool.unused_resources() == 0U); { auto o1 = pool.allocate(); } // Heavy object is back in the pool assert(pool.unused_resources() == 1U);
In this case we use the default constructor of the
recycle::shared_pool
this will only work if the object in this
case heavy_object
is default constructible (i.e. has a constructor
which takes no arguments). Internally the resource pool uses
std::make_shared
to allocate the object.
Example:
#include <recycle/shared_pool.hpp> #include <memory> struct heavy_object { heavy_object(std::size_t size); // ... some expensive resource }; auto make = []()->std::shared_ptr<heavy_object> { return std::make_shared<heavy_object>(300000U); }; recycle::shared_pool<heavy_object> pool(make); auto o1 = pool.allocate();
In this case we provide a custom allocator function which takes no
arguments and returns a std::shared_ptr
.
When recycling objects it is sometimes necessary to ensure that certain clean-up operations are performed before objects get stored in the pool. This can be open file handles etc. which should be closed. We cannot rely on the destructor for this when using a resource pool.
To support this the recycle::shared_pool
support a custom
recycle function which will be called right before an object is about
to go back into the pool.
Example:
#include <recycle/shared_pool.hpp> #include <memory> struct heavy_object { heavy_object(std::size_t size); // ... some expensive resource }; auto make = []()->std::shared_ptr<heavy_object> { return std::make_shared<heavy_object>(300000U); }; auto recycle = [](std::shared_ptr<heavy_object> o) { o->close_sockets(); }; recycle::shared_pool<heavy_object> pool(make, recycle); { auto o1 = pool.allocate(); // As we exit the scope here recycle will be called // with o1 as argument. }
Since the free lunch is over we want to make sure that the resource pool is thread safe.
This can be achieved by specifying a lock policy (we were inspired by the flyweight library in Boost).
Example:
#include <recycle/shared_pool.hpp> #include <mutex> #include <thread> struct heavy_object { // ... some expensive resource }; struct lock_policy { using mutex_type = std::mutex; using lock_type = std::lock_guard<mutex_type>; }; recycle::shared_pool<heavy_object, lock_policy> pool; // Lambda the threads will execute captures a reference to the pool // so they will all operate on the same pool concurrently auto run = [&pool]() { auto a1 = pool.allocate(); }; const std::size_t number_threads = 8; std::thread t[number_threads]; //Launch a group of threads for (std::size_t i = 0; i < number_threads; ++i) { t[i] = std::thread(run); } //Join the threads with the main thread for (std::size_t i = 0; i < number_threads; ++i) { t[i].join(); }
To depend on this project when using the CMake build system, add the following in your CMake build script:
add_subdirectory("/path/to/recycle" recycle) target_link_libraries(<my_target> steinwurf::recycle)
Where <my_target>
is replaced by your target.