[/ / Copyright (c) 2006-2013 Ion Gaztanaga / / Distributed under the Boost Software License, Version 1.0. (See accompanying / file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) /] [library Boost.Intrusive [quickbook 1.6] [authors [Krzikalla, Olaf], [Gaztanaga, Ion]] [copyright 2005 Olaf Krzikalla, 2006-2015 Ion Gaztanaga] [id intrusive] [dirname intrusive] [purpose Intrusive containers] [license Distributed under the Boost Software License, Version 1.0. (See accompanying file LICENSE_1_0.txt or copy at [@http://www.boost.org/LICENSE_1_0.txt]) ] ] [section:introduction Introduction] [section:introduction_presenting Presenting Boost.Intrusive] [*Boost.Intrusive] is a library presenting some intrusive containers to the world of C++. Intrusive containers are special containers that offer [link intrusive.performance better performance] and exception safety guarantees than non-intrusive containers (like STL containers). The performance benefits of intrusive containers makes them ideal as a building block to efficiently construct complex containers like multi-index containers or to design high performance code like memory allocation algorithms. While intrusive containers were and are widely used in C, they became more and more forgotten in C++ due to the presence of the standard containers which don't support intrusive techniques.[*Boost.Intrusive] wants to push intrusive containers usage encapsulating the implementation in STL-like interfaces. Hence anyone familiar with standard containers can easily use [*Boost.Intrusive]. [endsect] [section:introduction_building_intrusive Building Boost.Intrusive] There is no need to compile anything to use [*Boost.Intrusive], since it's a header only library. Just include your Boost header directory in your compiler include path. [endsect] [section:tested_compilers Tested compilers] [*Boost.Intrusive] has been tested on the following compilers/platforms: * Visual C++ >= 7.1. * GCC >= 4.1. [warning GCC < 4.3 and MSVC < 9.0 are deprecated and will be removed in the next version.] [endsect] [endsect] [section:intrusive_vs_nontrusive Intrusive and non-intrusive containers] [section:differences_intrusive_vs_nontrusive Differences between intrusive and non-intrusive containers] The main difference between intrusive containers and non-intrusive containers is that in C++ non-intrusive containers store [*copies] of values passed by the user. Containers use the `Allocator` template parameter to allocate the stored values: [c++] #include #include int main() { std::list myclass_list; MyClass myclass(...); myclass_list.push_back(myclass); //The stored object is different from the original object assert(&myclass != &myclass_list.front()); return 0; } To store the newly allocated copy of `myclass`, the container needs additional data: `std::list` usually allocates nodes that contain pointers to the next and previous node and the value itself. Something similar to: [c++] //A possible implementation of a std::list node class list_node { list_node *next; list_node *previous; MyClass value; }; On the other hand, an intrusive container does not store copies of passed objects, but it stores the objects themselves. The additional data needed to insert the object in the container must be provided by the object itself. For example, to insert `MyClass` in an intrusive container that implements a linked list, `MyClass` must contain the needed ['next] and ['previous] pointers: [c++] class MyClass { MyClass *next; MyClass *previous; //Other members... }; int main() { acme_intrusive_list list; MyClass myclass; list.push_back(myclass); //"myclass" object is stored in the list assert(&myclass == &list.front()); return 0; } As we can see, knowing which additional data the class should contain is not an easy task. [*Boost.Intrusive] offers several intrusive containers and an easy way to make user classes compatible with those containers. [endsect] [section:properties_of_intrusive Properties of Boost.Intrusive containers] Semantically, a [*Boost.Intrusive] container is similar to a STL container holding pointers to objects. That is, if you have an intrusive list holding objects of type `T`, then `std::list` would allow you to do quite the same operations (maintaining and navigating a set of objects of type T and types derived from it). A non-intrusive container has some limitations: * An object can only belong to one container: If you want to share an object between two containers, you either have to store multiple copies of those objects or you need to use containers of pointers: `std::list`. * The use of dynamic allocation to create copies of passed values can be a performance and size bottleneck in some applications. Normally, dynamic allocation imposes a size overhead for each allocation to store bookkeeping information and a synchronization to protected concurrent allocation from different threads. * Before C++11, only copies of objects could be stored in non-intrusive containers. Still copy or move constructors and copy or move assignment operators are required and non-copyable and non-movable objects can't be stored in some containers. In any case, [*new] objects have to be created inside the container using constructors and the same object can't be shared between two containers. * It's not possible to store a derived object in a STL-container while retaining its original type. Intrusive containers have some important advantages: * Operating with intrusive containers doesn't invoke any memory management at all. The time and size overhead associated with dynamic memory can be minimized. * The same object can be inserted in more than one container at the same time with a tiny overhead in the object size. * Iterating an Intrusive container needs less memory accesses than the semantically equivalent container of pointers: iteration is faster. * Intrusive containers offer better exception guarantees than non-intrusive containers. In some situations intrusive containers offer a no-throw guarantee that can't be achieved with non-intrusive containers. * The computation of an iterator to an element from a pointer or reference to that element is a constant time operation (computing the position of `T*` in a `std::list` has linear complexity). * Intrusive containers offer predictability when inserting and erasing objects since no memory management is done with intrusive containers. Memory management usually is not a predictable operation so complexity guarantees from non-intrusive containers are looser than the guarantees offered by intrusive containers. Intrusive containers have also downsides: * Each type stored in an intrusive container needs additional memory holding the maintenance information needed by the container. Hence, whenever a certain type will be stored in an intrusive container [*you have to change the definition of that type] appropriately. Although this task is easy with [*Boost.Intrusive], touching the definition of a type is sometimes a crucial issue. * In intrusive containers you don't store a copy of an object, [*but rather the original object is linked with other objects in the container]. Objects don't need copy-constructors or assignment operators to be stored in intrusive containers. But you have to take care of possible side effects, whenever you change the contents of an object (this is especially important for associative containers). * The user [*has to manage the lifetime of inserted objects] independently from the containers. * Again you have to be [*careful]: in contrast to STL containers [*it's easy to render an iterator invalid] without touching the intrusive container directly, because the object can be disposed before is erased from the container. * [*Boost.Intrusive] containers are [*non-copyable and non-assignable]. Since intrusive containers don't have allocation capabilities, these operations make no sense. However, swapping can be used to implement move capabilities. To ease the implementation of copy constructors and assignment operators of classes storing [*Boost.Intrusive] containers, [*Boost.Intrusive] offers special cloning functions. See [link intrusive.clone_from Cloning Boost.Intrusive containers] section for more information. * Analyzing the thread safety of a program that uses containers is harder with intrusive containers, because the container might be modified indirectly without an explicit call to a container member. [table Summary of intrusive containers advantages and disadvantages [[Issue] [Intrusive] [Non-intrusive]] [[Memory management] [External] [Internal through allocator]] [[Insertion/Erasure time] [Faster] [Slower]] [[Memory locality] [Better] [Worse]] [[Can insert the same object in more than one container] [Yes] [No]] [[Exception guarantees] [Better] [Worse]] [[Computation of iterator from value] [Constant] [Non-constant]] [[Insertion/erasure predictability] [High] [Low]] [[Memory use] [Minimal] [More than minimal]] [[Insert objects by value retaining polymorphic behavior] [Yes] [No (slicing)]] [[User must modify the definition of the values to insert] [Yes] [No]] [[Containers are copyable] [No] [Yes]] [[Inserted object's lifetime managed by] [User (more complex)] [Container (less complex)]] [[Container invariants can be broken without using the container] [Easier] [Harder (only with containers of pointers)]] [[Thread-safety analysis] [Harder] [Easier]] ] For a performance comparison between Intrusive and Non-intrusive containers see [link intrusive.performance Performance] section. [endsect] [endsect] [section:usage How to use Boost.Intrusive] If you plan to insert a class in an intrusive container, you have to make some decisions influencing the class definition itself. Each class that will be used in an intrusive container needs some appropriate data members storing the information needed by the container. We will take a simple intrusive container, the intrusive list ([classref boost::intrusive::list boost::intrusive::list]), for the following examples, but all [*Boost.Intrusive] containers are very similar. To compile the example using [classref boost::intrusive::list boost::intrusive::list], just include: [c++] #include Every class to be inserted in an intrusive container, needs to contain a hook that will offer the necessary data and resources to be insertable in the container. With [*Boost.Intrusive] you just choose the hook to be a public base class or a public member of the class to be inserted. [*Boost.Intrusive] also offers more flexible hooks for advanced users, as explained in the chapter [link intrusive.function_hooks Using function hooks], but usually base or member hooks are good enough for most users. [section:usage_base_hook Using base hooks] For [classref boost::intrusive::list list], you can publicly derive from [classref boost::intrusive::list_base_hook list_base_hook]. [c++] template class list_base_hook; The class can take several options. [*Boost.Intrusive] classes receive arguments in the form `option_name`. You can specify the following options: * [*`tag`]: this argument serves as a tag, so you can derive from more than one [classref boost::intrusive::list_base_hook list_base_hook] and hence put an object in multiple intrusive lists at the same time. An incomplete type can serve as a tag. If you specify two base hooks, you [*must] specify a different tag for each one. Example: `list_base_hook< tag >`. If no tag is specified a default one will be used (more on default tags later). * [*`link_mode`]: The second template argument controls the linking policy. [*Boost.Intrusive] currently supports 3 modes: `normal_link`, `safe_link` and `auto_unlink`. By default, `safe_link` mode is used. More about these in sections [link intrusive.safe_hook Safe hooks] and [link intrusive.auto_unlink_hooks Auto-unlink hooks]. Example: `list_base_hook< link_mode >` * [*`void_pointer`]: this option is the pointer type to be used internally in the hook. The default value is `void *`, which means that raw pointers will be used in the hook. More about this in the section titled [link intrusive.using_smart_pointers Using smart pointers with Boost.Intrusive containers]. Example: `list_base_hook< void_pointer< my_smart_ptr >` For the following examples, let's forget the options and use the default values: [c++] #include using namespace boost::intrusive; class Foo //Base hook with default tag, raw pointers and safe_link mode : public list_base_hook<> { /**/ }; After that, we can define the intrusive list: [c++] template class list; `list` receives the type to be inserted in the container (`T`) as the first parameter and optionally, the user can specify options. We have 3 option types: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: All these options specify the relationship between the type `T` to be inserted in the list and the hook (since we can have several hooks in the same `T` type). `member_hook` will be explained a bit later and `value_traits` will be explained in the [link intrusive.value_traits Containers with custom ValueTraits] section. [*If no option is specified, the container will be configured to use the base hook with the default tag]. Some options configured for the hook (the type of the pointers, link mode, etc.) will be propagated to the container. * [*`constant_time_size`]: Specifies if a constant time `size()` function is demanded for the container. This will instruct the intrusive container to store an additional member to keep track of the current size of the container. By default, constant-time size is activated. * [*`size_type`]: Specifies an unsigned type that can hold the size of the container. This type will be the type returned by `list.size()` and the type stored in the intrusive container if `constant_time_size` is requested. The user normally will not need to change this type, but some containers can have a `size_type` that might be different from `std::size_t` (for example, STL-like containers use the `size_type` defined by their allocator). [*Boost.Intrusive] can be used to implement such containers specifying the type of the size. By default the type is `std::size_t`. Example of a constant-time size intrusive list that will store Foo objects, using the base hook with the default tag: [c++] typedef list FooList; Example of an intrusive list with non constant-time size that will store Foo objects: [c++] typedef list > FooList; Remember that the user must specify the base hook in the container declaration if the base hook has no default tag, because that usually means that the type has more than one base hook, and a container shall know which hook will be using: [c++] #include using namespace boost::intrusive; struct my_tag1; struct my_tag2; typedef list_base_hook< tag > BaseHook; typedef list_base_hook< tag > BaseHook2; class Foo : public BaseHook, public BaseHook2 { /**/ }; typedef list< Foo, base_hook > FooList; typedef list< Foo, base_hook > FooList2; Once the list is defined, we can use it: [c++] //An object to be inserted in the list Foo foo_object; FooList list; list.push_back(object); assert(&list.front() == &foo_object); [endsect] [section:usage_member_hook Using member hooks] Sometimes an 'is-a' relationship between list hooks and the list value types is not desirable. In this case, using a member hook as a data member instead of 'disturbing' the hierarchy might be the right way: you can add a public data member `list_member_hook<...>` to your class. This class can be configured with the same options as `list_base_hook` except the option `tag`: [c++] template class list_member_hook; Example: [c++] #include class Foo { public: list_member_hook<> hook_; //... }; When member hooks are used, the `member_hook` option is used to configure the list: [c++] //This option will configure "list" to use the member hook typedef member_hook, &Foo::hook_> MemberHookOption; //This list will use the member hook typedef list FooList; Now we can use the container: [c++] //An object to be inserted in the list Foo foo_object; FooList list; list.push_back(object); assert(&list.front() == &foo_object); [endsect] However, member hooks have some implementation limitations: If there is a virtual inheritance relationship between the parent and the member hook, then the distance between the parent and the hook is not a compile-time fixed value so obtaining the address of the parent from the member hook is not possible without reverse engineering compiler produced RTTI. Apart from this, the non-standard pointer to member implementation for classes with complex inheritance relationships in MSVC ABI compatible-compilers is not supported by member hooks since it also depends on compiler-produced RTTI information. [section:usage_both_hooks Using both hooks] You can insert the same object in several intrusive containers at the same time, using one hook per container. This is a full example using base and member hooks: [import ../example/doc_how_to_use.cpp] [doc_how_to_use_code] [endsect] [section:usage_lifetime Object lifetime] Even if the interface of [classref boost::intrusive::list list] is similar to `std::list`, its usage is a bit different: You always have to keep in mind that you directly store objects in intrusive containers, not copies. The lifetime of a stored object is not bound to or managed by the container: * When the container gets destroyed before the object, the object is not destroyed, so you have to be careful to avoid resource leaks. * When the object is destroyed before the container, your program is likely to crash, because the container contains a pointer to an non-existing object. [endsect] [endsect] [section:usage_when When to use?] Intrusive containers can be used for highly optimized algorithms, where speed is a crucial issue and: * additional memory management should be avoided. * the programmer needs to efficiently track the construction and destruction of objects. * exception safety, especially the no-throw guarantee, is needed. * the computation of an iterator to an element from a pointer or reference to that element should be a constant time operation. * it's important to achieve a well-known worst-time system response. * localization of data (e.g. for cache hit optimization) leads to measurable effects. The last point is important if you have a lot of containers over a set of elements. E.g. if you have a vector of objects (say, `std::vector`), and you also have a list storing a subset of those objects (`std::list`), then operating on an Object from the list iterator (`std::list::iterator`) requires two steps: * Access from the iterator (usually on the stack) to the list node storing a pointer to `Object`. * Access from the pointer to `Object` to the Object stored in the vector. While the objects themselves are tightly packed in the memory of the vector (a vector's memory is guaranteed to be contiguous), and form something like a data block, list nodes may be dispersed in the heap memory. Hence depending on your system you might get a lot of cache misses. The same doesn't hold for an intrusive list. Indeed, dereferencing an iterator from an intrusive list is performed in the same two steps as described above. But the list node is already embedded in the Object, so the memory is directly tracked from the iterator to the Object. It's also possible to use intrusive containers when the objects to be stored can have different or unknown size. This allows storing base and derived objects in the same container, as shown in the following example: [import ../example/doc_window.cpp] [doc_window_code] Due to certain properties of intrusive containers they are often more difficult to use than their STL-counterparts. That's why you should avoid them in public interfaces of libraries. Classes to be stored in intrusive containers must change their implementation to store the hook and this is not always possible or desirable. [endsect] [section:concepts_summary Concept summary] Here is a small summary of the basic concepts that will be used in the following chapters: [variablelist Brief Concepts Summary [[Node Algorithms][A class containing typedefs and static functions that define basic operations that can be applied to a group of `node`s. It's independent from the node definition and configured using a NodeTraits template parameter that describes the node.]] [[Node Traits][A class that stores basic information and operations to insert a node into a group of nodes.]] [[Hook][A class that a user must add as a base class or as a member to make the user class compatible with intrusive containers. A Hook encapsulates a `node`]] [[Intrusive Container][A class that stores user classes that have the needed hooks. It takes a ValueTraits template parameter as configuration information.]] [[Semi-Intrusive Container][Similar to an intrusive container but a semi-intrusive container needs additional memory (e.g. an auxiliary array) to work.]] [[Value Traits][A class containing typedefs and operations to obtain the node to be used by Node Algorithms from the user class and the inverse.]] ] [endsect] [section:presenting_containers Presenting Boost.Intrusive containers] [*Boost.Intrusive] offers a wide range of intrusive containers: * [*slist]: An intrusive singly linked list. The size overhead is very small for user classes (usually the size of one pointer) but many operations have linear time complexity, so the user must be careful if he wants to avoid performance problems. * [*list]: A `std::list` like intrusive linked list. The size overhead is quite small for user classes (usually the size of two pointers). Many operations have constant time complexity. * [*set/multiset/rbtree]: `std::set/std::multiset` like intrusive associative containers based on red-black trees. The size overhead is moderate for user classes (usually the size of three pointers). Many operations have logarithmic time complexity. * [*avl_set/avl_multiset/avltree]: A `std::set/std::multiset` like intrusive associative containers based on AVL trees. The size overhead is moderate for user classes (usually the size of three pointers). Many operations have logarithmic time complexity. * [*splay_set/splay_multiset/splaytree]: `std::set/std::multiset` like intrusive associative containers based on splay trees. Splay trees have no constant operations, but they have some interesting caching properties. The size overhead is moderate for user classes (usually the size of three pointers). Many operations have logarithmic time complexity. * [*sg_set/sg_multiset/sgtree]: A `std::set/std::multiset` like intrusive associative containers based on scapegoat trees. Scapegoat can be configured with the desired balance factor to achieve the desired rebalancing frequency/search time compromise. The size overhead is moderate for user classes (usually the size of three pointers). Many operations have logarithmic time complexity. [*Boost.Intrusive] also offers semi-intrusive containers: * [*unordered_set/unordered_multiset]: `std::tr1::unordered_set/std::tr1::unordered_multiset` like intrusive unordered associative containers. The size overhead is moderate for user classes (an average of two pointers per element). Many operations have amortized constant time complexity. Most of these intrusive containers can be configured with constant or linear time size: * [*Linear time size]: The intrusive container doesn't hold a size member that is updated with every insertion/erasure. This implies that the `size()` function doesn't have constant time complexity. On the other hand, the container is smaller, and some operations, like `splice()` taking a range of iterators in linked lists, have constant time complexity instead of linear complexity. * [*Constant time size]: The intrusive container holds a size member that is updated with every insertion/erasure. This implies that the `size()` function has constant time complexity. On the other hand, increases the size of the container, and some operations, like `splice()` taking a range of iterators, have linear time complexity in linked lists. To make user classes compatible with these intrusive containers [*Boost.Intrusive] offers two types of hooks for each container type: * [*Base hook]: The hook is stored as a public base class of the user class. * [*Member hook]: The hook is stored as a public member of the user class. Apart from that, [*Boost.Intrusive] offers additional features: * [*Safe mode hooks]: Hook constructor initializes the internal `node` to a well-known safe state and intrusive containers check that state before inserting a value in the container using that hook. When erasing an element from the container, the container puts the `node` of the hook in the safe state again. This allows a safer use mode and it can be used to detect programming errors. It implies a slight performance overhead in some operations and can convert some constant time operations to linear time operations. * [*Auto-unlink hooks]: The hook destructor removes the object from the container automatically and the user can safely unlink the object from the container without referring to the container. * [*Non-raw pointers]: If the user wants to use smart pointers instead of raw pointers, [*Boost.Intrusive] hooks can be configured to use any type of pointer. This configuration information is also transmitted to the containers, so all the internal pointers used by intrusive containers configured with these hooks will be smart pointers. As an example, [*Boost.Interprocess] defines a smart pointer compatible with shared memory, called `offset_ptr`. [*Boost.Intrusive] can be configured to use this smart pointer to allow shared memory intrusive containers. [endsect] [section:safe_hook Safe hooks] [section:features Features of the safe mode] [*Boost.Intrusive] hooks can be configured to operate in safe-link mode. The safe mode is activated by default, but it can be also explicitly activated: [c++] //Configuring the safe mode explicitly class Foo : public list_base_hook< link_mode > {}; With the safe mode the user can detect if the object is actually inserted in a container without any external reference. Let's review the basic features of the safe mode: * Hook's constructor puts the hook in a well-known default state. * Hook's destructor checks if the hook is in the well-known default state. If not, an assertion is raised. * Every time an object is inserted in the intrusive container, the container checks if the hook is in the well-known default state. If not, an assertion is raised. * Every time an object is being erased from the intrusive container, the container puts the erased object in the well-known default state. With these features, without any external reference the user can know if the object has been inserted in a container by calling the `is_linked()` member function. If the object is not actually inserted in a container, the hook is in the default state, and if it is inserted in a container, the hook is not in the default state. [endsect] [section:configuring Configuring safe-mode assertions] By default, all safe-mode assertions raised by [*Boost-Intrusive] hooks and containers in are implemented using `BOOST_ASSERT`, which can be configured by the user. See [@http://www.boost.org/libs/utility/assert.html] for more information about `BOOST_ASSERT`. `BOOST_ASSERT` is globally configured, so the user might want to redefine intrusive safe-mode assertions without modifying the global `BOOST_ASSERT`. This can be achieved redefining the following macros: * `BOOST_INTRUSIVE_SAFE_HOOK_DEFAULT_ASSERT`: This assertion will be used in insertion functions of the intrusive containers to check that the hook of the value to be inserted is default constructed. * `BOOST_INTRUSIVE_SAFE_HOOK_DESTRUCTOR_ASSERT`: This assertion will be used in hooks' destructors to check that the hook is in a default state. If any of these macros is not redefined, the assertion will default to `BOOST_ASSERT`. If `BOOST_INTRUSIVE_SAFE_HOOK_DEFAULT_ASSERT` or `BOOST_INTRUSIVE_SAFE_HOOK_DESTRUCTOR_ASSERT` is defined and the programmer needs to include a file to configure that assertion, it can define `BOOST_INTRUSIVE_SAFE_HOOK_DESTRUCTOR_ASSERT_INCLUDE` or `BOOST_INTRUSIVE_SAFE_HOOK_DEFAULT_ASSERT_INCLUDE` with the name of the file to include: [c++] #define BOOST_INTRUSIVE_SAFE_HOOK_DESTRUCTOR_ASSERT MYASSERT #define BOOST_INTRUSIVE_SAFE_HOOK_DESTRUCTOR_ASSERT_INCLUDE [endsect] [endsect] [section:auto_unlink_hooks Auto-unlink hooks] [section:auto_unlink_hooks_what What's an auto-unlink hook?] [*Boost.Intrusive] offers additional hooks with unique features: * When the destructor of the hook is called, the hook checks if the node is inserted in a container. If so, the hook removes the node from the container. * The hook has a member function called `unlink()` that can be used to unlink the node from the container at any time, without having any reference to the container, if the user wants to do so. These hooks have exactly the same size overhead as their analog non auto-unlinking hooks, but they have a restriction: they can only be used with [link intrusive.presenting_containers non-constant time size containers]. There is a reason for this: * Auto-unlink hooks don't store any reference to the container where they are inserted. * Only containers with non constant-time `size()` allow removing an object from the container without referring to the container. This auto-unlink feature is useful in certain applications but it must be used [*very carefully]: * If several threads are using the same container the destructor of the auto-unlink hook will be called without any thread synchronization so removing the object is thread-unsafe. * Container contents change silently without modifying the container directly. This can lead to surprising effects. These auto-unlink hooks have also safe-mode properties: * Hooks' constructors put the hook in a well-known default state. * Every time an object is inserted in the intrusive container, the container checks if the hook is in the well-known default state. If not, an assertion is raised. * Every time an object is erased from an intrusive container, the container puts the erased object in the well-known default state. [endsect] [section:auto_unlink_hooks_example Auto-unlink hook example] Let's see an example of an auto-unlink hook: [import ../example/doc_auto_unlink.cpp] [doc_auto_unlink_code] [endsect] [section:auto_unlink_and_constant_time Auto-unlink hooks and containers with constant-time `size()`] As explained, [*Boost.Intrusive] auto-unlink hooks are incompatible with containers that have constant-time `size()`, so if you try to define such container with an auto-unlink hook's value_traits, you will get a static assertion: [c++] #include using boost::intrusive; struct MyTag; class MyClass : public list_base_hook< link_mode > {/**/}; list > bad_list; int main() { bad_list list; return 0; } leads to an error similar to: [pre error : use of undefined type 'boost::STATIC_ASSERTION_FAILURE' ] Pointing to code like this: [c++] //Constant-time size is incompatible with auto-unlink hooks! BOOST_STATIC_ASSERT(!(constant_time_size && ((int)value_traits::link_mode == (int)auto_unlink))); This way, there is no way to compile a program if you try to use auto-unlink hooks in constant-time size containers. [endsect] [endsect] [section:slist Intrusive singly linked list: slist] [classref boost::intrusive::slist slist] is the simplest intrusive container of [*Boost.Intrusive]: a singly linked list. The memory overhead it imposes is 1 pointer per node. The size of an empty, non constant-time size [classref boost::intrusive::slist slist] is the size of 1 pointer. This lightweight memory overhead comes with drawbacks, though: many operations have linear time complexity, even some that usually are constant time, like [classref boost::intrusive::slist::swap swap]. [classref boost::intrusive::slist slist] only provides forward iterators. For most cases, a doubly linked list is preferable because it offers more constant-time functions with a slightly bigger size overhead. However, for some applications like constructing more elaborate containers, singly linked lists are essential because of their low size overhead. [section:slist_hooks slist hooks] Like the rest of [*Boost.Intrusive] containers, [classref boost::intrusive::slist slist] has two hook types: [c++] template class slist_base_hook; * [classref boost::intrusive::slist_base_hook slist_base_hook]: the user class derives publicly from [classref boost::intrusive::slist_base_hook slist_base_hook] to make it [classref boost::intrusive::slist slist]-compatible. [c++] template class slist_member_hook; * [classref boost::intrusive::slist_member_hook slist_member_hook]: the user class contains a public [classref boost::intrusive::slist_member_hook slist_member_hook] to make it [classref boost::intrusive::slist slist]-compatible. [classref boost::intrusive::slist_base_hook slist_base_hook] and [classref boost::intrusive::slist_member_hook slist_member_hook] receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`tag`] (for base hooks only): This argument serves as a tag, so you can derive from more than one slist hook. Default: `tag`. * [*`link_mode`]: The linking policy. Default: `link_mode`. * [*`void_pointer`]: The pointer type to be used internally in the hook and propagated to the container. Default: `void_pointer`. [endsect] [section:slist_container slist container] [c++] template class slist; [classref boost::intrusive::slist slist] receives the options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`constant_time_size`]: To activate the constant-time `size()` operation. Default: `constant_time_size` * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type`. [classref boost::intrusive::slist slist] can receive additional options: * [*`linear`]: the singly linked list is implemented as a null-terminated list instead of a circular list. This allows `O(1)` swap, but losses some operations like `container_from_end_iterator`. * [*`cache_last`]: `slist` also stores a pointer to the last element of the singly linked list. This allows `O(1)` swap, `splice_after(iterator, slist &)` and makes the list offer new functions like `push_back(reference)` and `back()`. Logically, the size an empty list is increased in `sizeof(void_pointer)` and the cached last node pointer must be updated in every operation, and that might incur in a slight performance impact. `auto_unlink` hooks are not usable if `linear` and/or `cache_last` options are used. If `auto_unlink` hooks are used and those options are specified, a static assertion will be raised. [endsect] [section:slist_example Example] Now let's see a small example using both hooks: [import ../example/doc_slist.cpp] [doc_slist_code] [endsect] [endsect] [section:list Intrusive doubly linked list: list] [classref boost::intrusive::list list] is a doubly linked list. The memory overhead it imposes is 2 pointers per node. An empty, non constant-time size [classref boost::intrusive::list list] also has the size of 2 pointers. [classref boost::intrusive::list list] has many more constant-time operations than [classref boost::intrusive::slist slist] and provides a bidirectional iterator. It is recommended to use [classref boost::intrusive::list list] instead of [classref boost::intrusive::slist slist] if the size overhead is acceptable: [section:list_hooks list hooks] Like the rest of [*Boost.Intrusive] containers, [classref boost::intrusive::list list] has two hook types: [c++] template class list_base_hook; * [classref boost::intrusive::list_base_hook list_base_hook]: the user class derives publicly from [classref boost::intrusive::list_base_hook list_base_hook] to make it [classref boost::intrusive::list list]-compatible. [c++] template class list_member_hook; * [classref boost::intrusive::list_member_hook list_member_hook]: the user class contains a public [classref boost::intrusive::list_member_hook list_member_hook] to make it [classref boost::intrusive::list list]-compatible. [classref boost::intrusive::list_base_hook list_base_hook] and [classref boost::intrusive::list_member_hook list_member_hook] receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`tag`] (for base hooks only): This argument serves as a tag, so you can derive from more than one list hook. Default: `tag`. * [*`link_mode`]: The linking policy. Default: `link_mode`. * [*`void_pointer`]: The pointer type to be used internally in the hook and propagated to the container. Default: `void_pointer`. [endsect] [section:list_container list container] [c++] template class list; [classref boost::intrusive::list list] receives the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`constant_time_size`]: To activate the constant-time `size()` operation. Default: `constant_time_size` * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type` [endsect] [section:list_example Example] Now let's see a small example using both hooks: [import ../example/doc_list.cpp] [doc_list_code] [endsect] [endsect] [section:set_multiset Intrusive associative containers: set, multiset, rbtree] [*Boost.Intrusive] also offers associative containers that can be very useful when creating more complex associative containers, like containers maintaining one or more indices with different sorting semantics. Boost.Intrusive associative containers, like most STL associative container implementations are based on red-black trees. The memory overhead of these containers is usually 3 pointers and a bit (with alignment issues, this means 3 pointers and an integer). This size can be reduced to 3 pointers if pointers have even alignment (which is usually true in most systems). An empty, non constant-time size [classref boost::intrusive::set set], [classref boost::intrusive::multiset multiset] or [classref boost::intrusive::rbtree rbtree] has also the size of 3 pointers and an integer (3 pointers when optimized for size). These containers have logarithmic complexity in many operations like searches, insertions, erasures, etc. [classref boost::intrusive::set set] and [classref boost::intrusive::multiset multiset] are the intrusive equivalents of standard `std::set` and `std::multiset` containers. [classref boost::intrusive::rbtree rbtree] is a superset of [classref boost::intrusive::set set] and [classref boost::intrusive::multiset multiset] containers that offers functions to insert unique and multiple keys. [section:set_multiset_hooks set, multiset and rbtree hooks] [classref boost::intrusive::set set], [classref boost::intrusive::multiset multiset] and [classref boost::intrusive::rbtree rbtree] share the same hooks. This is an advantage, because the same user type can be inserted first in a [classref boost::intrusive::multiset multiset] and after that in [classref boost::intrusive::set set] without changing the definition of the user class. [c++] template class set_base_hook; * [classref boost::intrusive::set_base_hook set_base_hook]: the user class derives publicly from [classref boost::intrusive::set_base_hook set_base_hook] to make it [classref boost::intrusive::set set]/[classref boost::intrusive::multiset multiset]-compatible. [c++] template class set_member_hook; * [classref boost::intrusive::set_member_hook set_member_hook]: the user class contains a public [classref boost::intrusive::set_member_hook set_member_hook] to make it [classref boost::intrusive::set set]/[classref boost::intrusive::multiset multiset]-compatible. [classref boost::intrusive::set_base_hook set_base_hook] and [classref boost::intrusive::set_member_hook set_member_hook] receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive] plus a size optimization option: * [*`tag`] (for base hooks only): This argument serves as a tag, so you can derive from more than one base hook. Default: `tag`. * [*`link_mode`]: The linking policy. Default: `link_mode`. * [*`void_pointer`]: The pointer type to be used internally in the hook and propagated to the container. Default: `void_pointer`. * [*`optimize_size`]: The hook will be optimized for size instead of speed. The hook will embed the color bit of the red-black tree node in the parent pointer if pointer alignment is even. In some platforms, optimizing the size might reduce speed performance a bit since masking operations will be needed to access parent pointer and color attributes, in other platforms this option improves performance due to improved memory locality. Default: `optimize_size`. [endsect] [section:set_multiset_containers set, multiset and rbtree containers] [c++] template class set; template class multiset; template class rbtree; These containers receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`constant_time_size`]: To activate the constant-time `size()` operation. Default: `constant_time_size` * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type` And they also can receive an additional option: * [*`compare`]: Comparison function for the objects to be inserted in containers. The comparison functor must induce a strict weak ordering. Default: `compare< std::less >` * [*`key_of_value`]: A function object that will define the `key_type` of the value type to be stored. This type will allow a map-like interface. See [link intrusive.map_multimap Map and multimap-like interface with set and multiset] for details. Default: `key_type` is equal to `value_type` (set-like interface). [endsect] [section:set_multiset_example Example] Now let's see a small example using both hooks and both containers: [import ../example/doc_set.cpp] [doc_set_code] [endsect] [endsect] [section:unordered_set_unordered_multiset Semi-Intrusive unordered associative containers: unordered_set, unordered_multiset] [*Boost.Intrusive] also offers hashed containers that can be very useful to implement fast-lookup containers. These containers ([classref boost::intrusive::unordered_set unordered_set] and [classref boost::intrusive::unordered_multiset unordered_multiset]) are semi-intrusive containers: they need additional memory apart from the hook stored in the `value_type`. This additional memory must be passed in the constructor of the container. Unlike C++ TR1 unordered associative containers (which are also hashed containers), the contents of these semi-intrusive containers are not rehashed to maintain a load factor: that would require memory management and intrusive containers don't implement any memory management at all. However, the user can request an explicit rehashing passing a new bucket array. This also offers an additional guarantee over TR1 unordered associative containers: [*iterators are not invalidated when inserting an element] in the container. As with TR1 unordered associative containers, rehashing invalidates iterators, changes ordering between elements and changes which buckets elements appear in, but does not invalidate pointers or references to elements. Apart from expected hash and equality function objects, [*Boost.Intrusive] unordered associative containers' constructors take an argument specifying an auxiliary bucket vector to be used by the container. [section:unordered_set_unordered_multiset_performance unordered_set and unordered_multiset performance notes] The size overhead for a hashed container is moderate: 1 pointer per value plus a bucket array per container. The size of an element of the bucket array is usually one pointer. To obtain a good performance hashed container, the bucket length is usually the same as the number of elements that the container contains, so a well-balanced hashed container (`bucket_count()` is equal to `size()` ) will have an equivalent overhead of two pointers per element. An empty, non constant-time size [classref boost::intrusive::unordered_set unordered_set] or [classref boost::intrusive::unordered_multiset unordered_multiset] has also the size of `bucket_count()` pointers. Insertions, erasures, and searches, have amortized constant-time complexity in hashed containers. However, some worst-case guarantees are linear. See [classref boost::intrusive::unordered_set unordered_set] or [classref boost::intrusive::unordered_multiset unordered_multiset] for complexity guarantees of each operation. [*Be careful with non constant-time size hashed containers]: some operations, like `empty()`, have linear complexity, unlike other [*Boost.Intrusive] containers. [endsect] [section:unordered_set_unordered_multiset_hooks unordered_set and unordered_multiset hooks] [classref boost::intrusive::unordered_set unordered_set] and [classref boost::intrusive::unordered_multiset unordered_multiset] share the same hooks. This is an advantage, because the same user type can be inserted first in a [classref boost::intrusive::unordered_multiset unordered_multiset] and after that in [classref boost::intrusive::unordered_set unordered_set] without changing the definition of the user class. [c++] template class unordered_set_base_hook; * [classref boost::intrusive::unordered_set_base_hook unordered_set_base_hook]: the user class derives publicly from [classref boost::intrusive::unordered_set_base_hook unordered_set_base_hook] to make it [classref boost::intrusive::unordered_set unordered_set]/[classref boost::intrusive::unordered_multiset unordered_multiset]-compatible. [c++] template class unordered_set_member_hook; * [classref boost::intrusive::unordered_set_member_hook unordered_set_member_hook]: the user class contains a public [classref boost::intrusive::unordered_set_member_hook unordered_set_member_hook] to make it [classref boost::intrusive::unordered_set unordered_set]/[classref boost::intrusive::unordered_multiset unordered_multiset]-compatible. [classref boost::intrusive::unordered_set_base_hook unordered_set_base_hook] and [classref boost::intrusive::unordered_set_member_hook unordered_set_member_hook] receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`tag`] (for base hooks only): This argument serves as a tag, so you can derive from more than one base hook. Default: `tag`. * [*`link_mode`]: The linking policy. Default: `link_mode`. * [*`void_pointer`]: The pointer type to be used internally in the hook and propagated to the container. Default: `void_pointer`. Apart from them, these hooks offer additional options: * [*`store_hash`]: This option reserves additional space in the hook to store the hash value of the object once it's introduced in the container. When this option is used, the unordered container will store the calculated hash value in the hook and rehashing operations won't need to recalculate the hash of the value. This option will improve the performance of unordered containers when rehashing is frequent or hashing the value is a slow operation. Default: `store_hash`. * [*`optimize_multikey`]: This option reserves additional space in the hook that will be used to group equal elements in unordered multisets, improving significantly the performance when many equal values are inserted in these containers. Default: `optimize_multikey`. [endsect] [section:unordered_set_unordered_multiset_containers unordered_set and unordered_multiset containers] [c++] template class unordered_set; template class unordered_multiset; As mentioned, unordered containers need an auxiliary array to work. [*Boost.Intrusive] unordered containers receive this auxiliary array packed in a type called `bucket_traits` (which can be also customized by a container option). All unordered containers receive a `bucket_traits` object in their constructors. The default `bucket_traits` class is initialized with a pointer to an array of buckets and its size: [c++] #include using namespace boost::intrusive; struct MyClass : public unordered_set_base_hook<> {}; typedef unordered_set::bucket_type bucket_type; typedef unordered_set::bucket_traits bucket_traits; int main() { bucket_type buckets[100]; unordered_set uset(bucket_traits(buckets, 100)); return 0; } Each hashed container needs [*its own bucket traits], that is, [*its own bucket vector]. Two hashed containers [*can't] share the same `bucket_type` elements. The bucket array [*must] be destroyed [*after] the container using it is destroyed, otherwise, the result is undefined. [classref boost::intrusive::unordered_set unordered_set] and [classref boost::intrusive::unordered_multiset unordered_multiset] receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`constant_time_size`]: To activate the constant-time `size()` operation. Default: `constant_time_size` * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type` And they also can receive additional options: * [*`equal`]: Equality function for the objects to be inserted in containers. Default: `equal< std::equal_to >` * [*`hash`]: Hash function to be used in the container. Default: `hash< boost::hash >` * [*`bucket_traits`]: A type that wraps the bucket vector to be used by the unordered container. Default: a type initialized by the address and size of a bucket array and stores both variables internally. * [*`power_2_buckets`]: The user guarantees that only bucket arrays with power of two length will be used. The container will then use masks instead of modulo operations to obtain the bucket number from the hash value. Masks are faster than modulo operations and for some applications modulo operations can impose a considerable overhead. In debug mode an assertion will be raised if the user provides a bucket length that is not power of two. Default: `power_2_buckets`. * [*`cache_begin`]: [*Note: this option is not compatible with `auto_unlink` hooks]. Due to its internal structure, finding the first element of an unordered container (`begin()` operation) is amortized constant-time. It's possible to speed up `begin()` and other operations related to it (like `clear()`) if the container caches internally the position of the first element. This imposes the overhead of one pointer to the size of the container. Default: `cache_begin`. * [*`compare_hash`]: [*Note: this option requires `store_hash` option in the hook]. When the comparison function is expensive, (e.g. strings with a long common predicate) sometimes (specially when the load factor is high or we have many equivalent elements in an [classref boost::intrusive::unordered_multiset unordered_multiset] and no `optimize_multikey<>` is activated in the hook) the equality function is a performance problem. Two equal values must have equal hashes, so comparing the hash values of two elements before using the comparison functor can speed up some implementations. * [*`incremental`]: Activates incremental hashing (also known as Linear Hashing). This option implies `power_2_buckets` and the container will require power of two buckets. For more information on incremental hashing, see [@http://en.wikipedia.org/wiki/Linear_hashing `Linear hash` on Wikipedia] Default: `incremental` * [*`key_of_value`]: A function object that will define the `key_type` of the value type to be stored. This type will allow a map-like interface. See [link intrusive.map_multimap Map and multimap-like interface with set and multiset] for details. Default: `key_type` is equal to `value_type` (set-like interface). [endsect] [section:unordered_set_unordered_multiset_example Example] Now let's see a small example using both hooks and both containers: [import ../example/doc_unordered_set.cpp] [doc_unordered_set_code] [endsect] [section:custom_bucket_traits Custom bucket traits] Instead of using the default `bucket_traits` class to store the bucket array, a user can define his own class to store the bucket array using the [*['bucket_traits<>]] option. A user-defined bucket-traits must fulfill the following interface: [c++] class my_bucket_traits { bucket_ptr bucket_begin(); const_bucket_ptr bucket_begin() const; std::size_t bucket_count() const; }; The following bucket traits just stores a pointer to the bucket array but the size is a compile-time constant. Note the use of the auxiliary [classref boost::intrusive::unordered_bucket unordered_bucket] and [classref boost::intrusive::unordered_bucket_ptr unordered_bucket_ptr] utilities to obtain the type of the bucket and its pointer before defining the unordered container: [import ../example/doc_bucket_traits.cpp] [doc_bucket_traits] [endsect] [endsect] [section:map_multimap Map and multimap-like interface for associative containers] Implementing map-like intrusive containers is not a trivial task as STL's `std::map` and `std::multimap` containers store copies of a `value_type` which is defined as `std::pair`. In order to reproduce this interface in [*Boost.Intrusive] it shall require that objects stored in the intrusive containers contain that `std::pair` member with `pair.first` hardcoded as the key part and `pair.second` hardcoded as the `mapped_type`, which is limiting and also not very useful in practice. Any intrusive associative container can be used like a map using [link intrusive.advanced_lookups_insertions advanced lookup and insertions] and the user can change the key type in each lookup/insertion check call. On the other hand, in many cases containers are indexed by a well-known key type, and the user is forced to write some repetitive code using advanced lookup and insertions. [*Boost.Intrusive] associative containers offer an alternative to build a useful map-like lookup interfaces without forcing users to define `value_type`s containing `std::pair`-like classes. The option is called [classref boost::intrusive::key_of_value]. If a user specifies that option when defining a `set/multiset` intrusive container, it specifies a function object that will tell the container which is the type of the ['key] that `value_type` holds and how to obtain it. This function object must be: * Lightweight to copy. * Default constructible (when the container constructor overload requires it). * It shall define: * A `type` member that defines the type of the key * A member function that returns the key derived a `value_type`, either by value or by const-reference. Let's see an example of how a set can be configured as a map indexed by an integer stored in the `value_type`: [import ../example/doc_map.cpp] [doc_map_code] [endsect] [section:avl_set_multiset Intrusive avl tree based associative containers: avl_set, avl_multiset and avltree] Similar to red-black trees, AVL trees are balanced binary trees. AVL trees are often compared with red-black trees because they support the same set of operations and because both take O(log n) time for basic operations. AVL trees are more rigidly balanced than Red-Black trees, leading to slower insertion and removal but faster retrieval, so AVL trees perform better than red-black trees for lookup-intensive applications. [*Boost.Intrusive] offers 3 containers based on avl trees: [classref boost::intrusive::avl_set avl_set], [classref boost::intrusive::avl_multiset avl_multiset] and [classref boost::intrusive::avltree avltree]. The first two are similar to [classref boost::intrusive::set set] or [classref boost::intrusive::multiset multiset] and the latter is a generalization that offers functions both to insert unique and multiple keys. The memory overhead of these containers with Boost.Intrusive hooks is usually 3 pointers and 2 bits (due to alignment, this usually means 3 pointers plus an integer). This size can be reduced to 3 pointers if pointers have 4 byte alignment (which is usually true in 32 bit systems). An empty, non constant-time size [classref boost::intrusive::avl_set avl_set], [classref boost::intrusive::avl_multiset avl_multiset] or [classref boost::intrusive::avltree avltree] also has a size of 3 pointers and an integer (3 pointers when optimized for size). [section:avl_set_multiset_hooks avl_set, avl_multiset and avltree hooks] [classref boost::intrusive::avl_set avl_set], [classref boost::intrusive::avl_multiset avl_multiset] and [classref boost::intrusive::avltree avltree] share the same hooks. [c++] template class avl_set_base_hook; * [classref boost::intrusive::avl_set_base_hook avl_set_base_hook]: the user class derives publicly from this class to make it compatible with avl tree based containers. [c++] template class avl_set_member_hook; * [classref boost::intrusive::set_member_hook set_member_hook]: the user class contains a public member of this class to make it compatible with avl tree based containers. [classref boost::intrusive::avl_set_base_hook avl_set_base_hook] and [classref boost::intrusive::avl_set_member_hook avl_set_member_hook] receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive] plus an option to optimize the size of the node: * [*`tag`] (for base hooks only): This argument serves as a tag, so you can derive from more than one base hook. Default: `tag`. * [*`link_mode`]: The linking policy. Default: `link_mode`. * [*`void_pointer`]: The pointer type to be used internally in the hook and propagated to the container. Default: `void_pointer`. * [*`optimize_size`]: The hook will be optimized for size instead of speed. The hook will embed the balance bits of the AVL tree node in the parent pointer if pointer alignment is multiple of 4. In some platforms, optimizing the size might reduce speed performance a bit since masking operations will be needed to access parent pointer and balance factor attributes, in other platforms this option improves performance due to improved memory locality. Default: `optimize_size`. [endsect] [section:set_multiset_containers avl_set, avl_multiset and avltree containers] [c++] template class avl_set; template class avl_multiset; template class avltree; These containers receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`constant_time_size`]: To activate the constant-time `size()` operation. Default: `constant_time_size` * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type` And they also can receive an additional option: * [*`compare`]: Comparison function for the objects to be inserted in containers. The comparison functor must induce a strict weak ordering. Default: `compare< std::less >` * [*`key_of_value`]: A function object that will define the `key_type` of the value type to be stored. This type will allow a map-like interface. See [link intrusive.map_multimap Map and multimap-like interface with set and multiset] for details. Default: `key_type` is equal to `value_type` (set-like interface). [endsect] [section:avl_set_multiset_example Example] Now let's see a small example using both hooks and [classref boost::intrusive::avl_set avl_set]/ [classref boost::intrusive::avl_multiset avl_multiset] containers: [import ../example/doc_avl_set.cpp] [doc_avl_set_code] [endsect] [endsect] [section:splay_set_multiset Intrusive splay tree based associative containers: splay_set, splay_multiset and , splay_tree] C++ associative containers are usually based on red-black tree implementations (e.g.: STL, Boost.Intrusive associative containers). However, there are other interesting data structures that offer some advantages (and also disadvantages). Splay trees are self-adjusting binary search trees used typically in caches, memory allocators and other applications, because splay trees have a "caching effect": recently accessed elements have better access times than elements accessed less frequently. For more information on splay trees see [@http://en.wikipedia.org/wiki/Splay_tree the corresponding Wikipedia entry]. [*Boost.Intrusive] offers 3 containers based on splay trees: [classref boost::intrusive::splay_set splay_set], [classref boost::intrusive::splay_multiset splay_multiset] and [classref boost::intrusive::splaytree splaytree]. The first two are similar to [classref boost::intrusive::set set] or [classref boost::intrusive::multiset multiset] and the latter is a generalization that offers functions both to insert unique and multiple keys. The memory overhead of these containers with Boost.Intrusive hooks is usually 3 pointers. An empty, non constant-time size splay container has also a size of 3 pointers. [section:splay_set_multiset_disadvantages Advantages and disadvantages of splay tree based containers] Splay tree based intrusive containers have logarithmic complexity in many operations like searches, insertions, erasures, etc., but if some elements are more frequently accessed than others, splay trees perform faster searches than equivalent balanced binary trees (such as red-black trees). The caching effect offered by splay trees comes with a cost: the tree must be rebalanced when an element is searched. To maintain const-correctness and thread-safety guarantees, this caching effect is not updated when const versions of search functions like `find()`, `lower_bound()`, `upper_bound()`, `equal_range()`, `count()`... are called. This means that using splay-tree based associative containers as drop-in replacements of [classref boost::intrusive::set set]/ [classref boost::intrusive::multiset multiset], specially for const search functions, might not result in desired performance improvements. If element searches are randomized, the tree will be continuously srebalanced without taking advantage of the cache effect, so splay trees can offer worse performance than other balanced trees for several search patterns. [*Boost.Intrusive] splay associative containers don't use their own hook types but plain Binary search tree hooks. See [link intrusive.bst_hooks Binary search tree hooks: bs_set_base_hook and bs_set_member_hook] section for more information about these hooks. [endsect] [section:set_multiset_containers splay_set, splay_multiset and splaytree containers] [c++] template class splay_set; template class splay_multiset; template class splaytree; These containers receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`constant_time_size`]: To activate the constant-time `size()` operation. Default: `constant_time_size` * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type` And they also can receive an additional option: * [*`compare`]: Comparison function for the objects to be inserted in containers. The comparison functor must induce a strict weak ordering. Default: `compare< std::less >` * [*`key_of_value`]: A function object that will define the `key_type` of the value type to be stored. This type will allow a map-like interface. See [link intrusive.map_multimap Map and multimap-like interface with set and multiset] for details. Default: `key_type` is equal to `value_type` (set-like interface). [endsect] [section:splay_set_multiset_example Example] Now let's see a small example using [classref boost::intrusive::splay_set splay_set]/ [classref boost::intrusive::splay_multiset splay_multiset] containers: [import ../example/doc_splay_set.cpp] [doc_splay_set_code] [endsect] [endsect] [section:sg_set_multiset Intrusive scapegoat tree based associative containers: sg_set, sg_multiset and sgtree] A scapegoat tree is a self-balancing binary search tree, that provides worst-case O(log n) lookup time, and O(log n) amortized insertion and deletion time. Unlike other self-balancing binary search trees that provide worst case O(log n) lookup time, scapegoat trees have no additional per-node overhead compared to a regular binary search tree. A binary search tree is said to be weight balanced if half the nodes are on the left of the root, and half on the right. An a-height-balanced tree is defined with defined with the following equation: [*['height(tree) <= log1/a(tree.size())]] * [*['a == 1]]: A tree forming a linked list is considered balanced. * [*['a == 0.5]]: Only a perfectly balanced binary is considered balanced. Scapegoat trees are loosely ['a-height-balanced] so: [*['height(tree) <= log1/a(tree.size()) + 1]] Scapegoat trees support any a between 0.5 and 1. If a is higher, the tree is rebalanced less often, obtaining quicker insertions but slower searches. Lower a values improve search times. Scapegoat-trees implemented in [*Boost.Intrusive] offer the possibility of [*changing a at run-time] taking advantage of the flexibility of scapegoat trees. For more information on scapegoat trees see [@http://en.wikipedia.org/wiki/Scapegoat_tree Wikipedia entry]. Scapegoat trees also have downsides: * They need additional storage of data on the root (the size of the tree, for example) to achieve logarithmic complexity operations so it's not possible to offer `auto_unlink` hooks. The size of an empty scapegoat tree is also considerably increased. * The operations needed to determine if the tree is unbalanced require floating-point operations like ['log1/a]. If the system has no floating point operations (like some embedded systems), scapegoat tree operations might become slow. [*Boost.Intrusive] offers 3 containers based on scapegoat trees: [classref boost::intrusive::sg_set sg_set], [classref boost::intrusive::sg_multiset sg_multiset] and [classref boost::intrusive::sgtree sgtree]. The first two are similar to [classref boost::intrusive::set set] or [classref boost::intrusive::multiset multiset] and the latter is a generalization that offers functions both to insert unique and multiple keys. The memory overhead of these containers with Boost.Intrusive hooks is 3 pointers. An empty, [classref boost::intrusive::sg_set sg_set], [classref boost::intrusive::sg_multiset sg_multiset] or [classref boost::intrusive::sgtree sgtree] has also the size of 3 pointers, two integers and two floating point values (equivalent to the size of 7 pointers on most systems). [*Boost.Intrusive] scapegoat associative containers don't use their own hook types but plain Binary search tree hooks. See [link intrusive.bst_hooks Binary search tree hooks: bs_set_base_hook and bs_set_member_hook] section for more information about these hooks. [section:sg_set_multiset_containers sg_set, sg_multiset and sgtree containers] [c++] template class sg_set; template class sg_multiset; template class sgtree; These containers receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type` And they also can receive additional options: * [*`compare`]: Comparison function for the objects to be inserted in containers. The comparison functor must induce a strict weak ordering. Default: `compare< std::less >` * [*`floating_point`]: When this option is deactivated, the scapegoat tree loses the ability to change the balance factor a at run-time, but the size of an empty container is reduced and no floating point operations are performed, normally increasing container performance. The fixed a factor that is used when this option is activated is ['1/sqrt(2) ~ 0,70711]. Default: `floating_point` * [*`key_of_value`]: A function object that will define the `key_type` of the value type to be stored. This type will allow a map-like interface. See [link intrusive.map_multimap Map and multimap-like interface with set and multiset] for details. Default: `key_type` is equal to `value_type` (set-like interface). [endsect] [section:sg_set_multiset_example Example] Now let's see a small example using binary search tree hooks and [classref boost::intrusive::sg_set sg_set]/ [classref boost::intrusive::sg_multiset sg_multiset] containers: [import ../example/doc_sg_set.cpp] [doc_sg_set_code] [endsect] [endsect] [section:treap_set_multiset Intrusive treap based associative containers: treap_set, treap_multiset and treap] The name ['treap] is a mixture of ['tree] and ['heap] indicating that Treaps exhibit the properties of both binary search trees and heaps. A treap is a binary search tree that orders the nodes by a key but also by a priority attribute. The nodes are ordered so that the keys form a binary search tree and the priorities obey the max heap order property. * If v is a left descendant of u, then key[v] < key[u]; * If v is a right descendant of u, then key[v] > key[u]; * If v is a child of u, then priority[v] <= priority[u]; If priorities are non-random, the tree will usually be unbalanced; this worse theoretical average-case behavior may be outweighed by better expected-case behavior, as the most important items will be near the root. This means most important objects will be retrieved faster than less important items and for items keys with equal keys most important objects will be found first. These properties are important for some applications. The priority comparison will be provided just like the key comparison, via a function object that will be stored in the intrusive container. This means that the priority can be stored in the value to be introduced in the treap or computed on flight (via hashing or similar). [*Boost.Intrusive] offers 3 containers based on treaps: [classref boost::intrusive::treap_set treap_set], [classref boost::intrusive::treap_multiset treap_multiset] and [classref boost::intrusive::treap treap]. The first two are similar to [classref boost::intrusive::set set] or [classref boost::intrusive::multiset multiset] and the latter is a generalization that offers functions both to insert unique and multiple keys. The memory overhead of these containers with Boost.Intrusive hooks is 3 pointers. An empty, [classref boost::intrusive::treap_set treap_set], [classref boost::intrusive::treap_multiset treap_multiset] or [classref boost::intrusive::treap treap] has also the size of 3 pointers and an integer (supposing empty function objects for key and priority comparison and constant-time size). [*Boost.Intrusive] treap associative containers don't use their own hook types but plain Binary search tree hooks. See [link intrusive.bst_hooks Binary search tree hooks: bs_set_base_hook and bs_set_member_hook] section for more information about these hooks. [section:treap_set_multiset_containers treap_set, treap_multiset and treap containers] [c++] template class treap_set; template class treap_multiset; template class treap; These containers receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`base_hook`] / [*`member_hook`] / [*`value_traits`]: To specify the hook type or value traits used to configure the container. (To learn about value traits go to the section [link intrusive.value_traits Containers with custom ValueTraits].) * [*`constant_time_size`]: To activate the constant-time `size()` operation. Default: `constant_time_size` * [*`size_type`]: To specify the type that will be used to store the size of the container. Default: `size_type` And they also can receive additional options: * [*`compare`]: Comparison function for the objects to be inserted in containers. The comparison functor must induce a strict weak ordering. Default: `compare< std::less >` * [*`priority_of_value`]: A function object that specifies the type of the priority (`priority_type`) of a treap container and an operator to obtain it from a value type. Default: `priority_type` is equal to `value_type` (set-like interface). * [*`priority`]: Priority Comparison function for the objects to be inserted in containers. The comparison functor must induce a strict weak ordering. Default: `priority< priority_compare >` * [*`key_of_value`]: A function object that will define the `key_type` of the value type to be stored. This type will allow a map-like interface. See [link intrusive.map_multimap Map and multimap-like interface with set and multiset] for details. Default: `key_type` is equal to `value_type` (set-like interface). The default `priority_compare` object function will call an unqualified function `priority_order` passing two constant `T` references as arguments and should return true if the first argument has higher priority (it will be searched faster), inducing strict weak ordering. The function will be found using ADL lookup so that the user just needs to define a `priority_order` function in the same namespace as the class: [c++] struct MyType { friend bool priority_order(const MyType &a, const MyType &b) {...} }; or namespace mytype { struct MyType{ ... }; bool priority_order(const MyType &a, const MyType &b) {...} } //namespace mytype { [endsect] [section:treap_set_exceptions Exception safety of treap-based intrusive containers] In general, intrusive containers offer strong safety guarantees, but treap containers must deal with two possibly throwing functors (one for value ordering, another for priority ordering). Moreover, treap erasure operations require rotations based on the priority order function and this issue degrades usual `erase(const_iterator)` no-throw guarantee. However, intrusive offers the strongest possible behaviour in these situations. In summary: * If the priority order functor does not throw, treap-based containers, offer exactly the same guarantees as other tree-based containers. * If the priority order functor throws, treap-based containers offer strong guarantee. [endsect] [section:treap_set_multiset_example Example] Now let's see a small example using binary search tree hooks and [classref boost::intrusive::treap_set treap_set]/ [classref boost::intrusive::treap_multiset treap_multiset] containers: [import ../example/doc_treap_set.cpp] [doc_treap_set_code] [endsect] [endsect] [section:bst_hooks Binary search tree hooks: bs_set_base_hook and bs_set_member_hook] Binary search tree hooks can be used with several tree-like containers that don't need any additional metadata for rebalancing operations. This has many advantages since binary search tree hooks can also be used to insert values in plain binary search tree, splay tree, scapegoat tree, and treap containers. [c++] template class bs_set_base_hook; * [classref boost::intrusive::bs_set_base_hook bs_set_base_hook]: the user class derives publicly from this class to make it compatible with the mentioned tree based containers. [c++] template class bs_set_member_hook; * [classref boost::intrusive::bs_set_member_hook bs_set_member_hook]: the user class contains a public member of this class to make it compatible with the mentioned tree based containers. [classref boost::intrusive::bs_set_base_hook bs_set_base_hook] and [classref boost::intrusive::bs_set_member_hook bs_set_member_hook] receive the same options explained in the section [link intrusive.usage How to use Boost.Intrusive]: * [*`tag`] (for base hooks only): This argument serves as a tag, so you can derive from more than one base hook. Default: `tag`. * [*`link_mode`]: The linking policy. Default: `link_mode`. * [*`void_pointer`]: The pointer type to be used internally in the hook and propagated to the container. Default: `void_pointer`. [endsect] [section:advanced_lookups_insertions Advanced lookup and insertion functions for associative containers] [section:advanced_lookups Advanced lookups] [*Boost.Intrusive] associative containers offer an interface similar to STL associative containers. However, STL's ordered and unordered simple associative containers (`std::set`, `std::multiset`, `std::unordered_set` and `std::unordered_multiset`) have some inefficiencies caused by the interface in several search, insertion or erasure functions (`equal_range`, `lower_bound`, `upper_bound`, `find`, `insert`, `erase`...): the user can only operate with `value_type` objects or (starting from C++11), [@http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2013/n3657.htm heterogeneous comparison lookups] which are not flexible enough as `key_compare` shall support the comparison between the provided key and `value_type`, which precludes the use of user-defined comparison objects that can partition the search in a compatible but advanced way. To solve these problems, [*Boost.Intrusive] containers offers functions where a key type different from `key_type` and a comparison object are provided by the user. This applies to: * equal_range * lower_bound * upper_bound * count * find * erase Requirements for such functions are: * For unordered container the provided comparison and hashing function with the given key shall induce the same hash and equivalence as `key_compare` and `hasher`. * For ordered associative containers, lookup and erasure functions, the container to be searched shall be partitioned in regards to the supplied comparison object and key. For more details, see [*Requires] clauses of such functions in the reference. [section:advanced_lookups_example Example] Imagine that the object to be searched is quite expensive to construct (called `Expensive` in the example): [import ../example/doc_assoc_optimized_code.cpp] [doc_assoc_optimized_code_normal_find] If "key" c-string is quite long `Expensive` has to construct a `std::string` using heap memory. Like `Expensive`, many times the only member taking part in ordering issues is just a small part of the class. E.g.: with `Expensive`, only the internal `std::string` is needed to compare the object. In both containers, if we call `get_from_set/get_from_unordered_set` in a loop, we might get a performance penalty, because we are forced to create a whole `Expensive` object to be able to find an equivalent one. Sometimes the problem is not only performance-related, as we [*might not have enough information to construct the object] but we might [*have enough information to find the object]. In this case, a name is enough to search `Expensive` objects in the container but constructing an `Expensive` object might require more information that the searcher might not have. To solve this, we can use the functions that take any type comparable with the value and a the 'compatible' comparison object (and hash, when the associative container is unordered) Let's see optimized search function: [doc_assoc_optimized_code_optimized_find] [endsect] [endsect] [section:advanced_insertions Advanced insertions] A similar issue happens with insertions in simple ordered and unordered associative containers with unique keys (`std::set` and `std::unordered_set`). In these containers, if a value is already present, the value to be inserted is discarded. With expensive values, if the value is already present, we can suffer efficiency problems. [classref boost::intrusive::set set] and [classref boost::intrusive::unordered_set unordered_set]-like containers have insertion functions (`insert_check`, `insert_unique_check`,...) to check efficiently, without constructing the value, if a value is present or not and if it's not present, a function to insert it immediately (`insert_commit`) without any further lookup. Requirements for functions that check the existence of such value are: * For unordered container the provided comparison and hashing function with the given key shall induce the same hash and equivalence as `key_compare` and `hasher`. * For ordered associative containers, the provided comparison function with the given key, shall induce the same strict weak order as `key_compare`. To sum up, `insert_check` is similar to a normal `insert` but: * `insert_check` can be used with arbitrary keys * if the insertion is possible (there is no equivalent value) `insert_check` collects all the needed information in an `insert_commit_data` structure, so that `insert_commit`: * [*does not execute] further comparisons * can be executed with [*constant-time complexity] * has [*no-throw guarantee]. These functions must be used with care, no other insertion or erasure must be executed between an `insert_check` and an `insert_commit` pair. Otherwise, the behaviour is undefined. See [classref boost::intrusive::set set] and [classref boost::intrusive::unordered_set unordered_set]-like containers' reference for more information about `insert_check` and `insert_commit`. With multiple ordered and unordered associative containers ([classref boost::intrusive::multiset multiset] and [classref boost::intrusive::unordered_multiset unordered_multiset]) there is no need for these advanced insertion functions, since insertions are always successful. [section:advanced_insertions_example Example] For example, using the same `Expensive` class, this function can be inefficient: [doc_assoc_optimized_code_normal_insert] If the object is already present, we are constructing an `Expensive` that will be discarded, and this is a waste of resources. Instead of that, let's use `insert_check` and `insert_commit` functions: [doc_assoc_optimized_code_optimized_insert] [endsect] [endsect] [section:positional_insertions Positional insertions] Some ordered associative containers offer low-level functions to bypass ordering checks and insert nodes directly in desired tree positions. These functions are provided for performance reasons when values to be inserted in the container are known to fulfill order (sets and multisets) and uniqueness (sets) invariants. A typical usage of these functions is when intrusive associative containers are used to build non-intrusive containers and the programmer wants to speed up assignments from other associative containers: if the ordering and uniqueness properties are the same, there is no need to waste time checking the position of each source value, because values are already ordered: back insertions will be much more efficient. [*Note:] These functions [*don't check preconditions] so they must used with care. They are low-level operations that [*will break container invariants if ordering and uniqueness preconditions are not assured by the caller.] Let's see an example: [import ../example/doc_positional_insertion.cpp] [doc_positional_insertion] [endsect] For more information about advanced lookup and insertion functions see associative containers' documentation (e.g. [classref boost::intrusive::set set], [classref boost::intrusive::multiset multiset], [classref boost::intrusive::unordered_set unordered_set] and [classref boost::intrusive::unordered_multiset unordered_multiset] references). [endsect] [section:erasing_and_disposing Erasing and disposing values from Boost.Intrusive containers] One of the most tedious tasks when using intrusive containers is the management of the erased elements. When using STL containers, the container itself unlinks and destroys the contained elements, but with intrusive containers, the user must explicitly destroy the object after erasing an element from the container. This makes STL-like functions erasing multiple objects unhelpful: the user can't destroy every erased element. For example, let's take the function `remove_if` from [classref boost::intrusive::list list]: [c++] template void remove_if(Pred pred); How can the user destroy the elements (say, using `operator delete`) that will be erased according to the predicate? [*Boost.Intrusive] containers offer additional functions that take a function object that will be called after the element has been erased from the container. For example, [classref boost::intrusive::list list] offers: [c++] template void remove_and_dispose_if(Pred pred, Disposer disposer) With this function the user can efficiently remove and destroy elements if the disposer function destroys an object: `remove_and_dispose_if` will call the "disposer" function object for every removed element. [classref boost::intrusive::list list] offers more functions taking a disposer function object as argument, like `erase_and_dispose`, `clear_and_dispose`, `remove_and_dispose`, etc. Note that the disposing function does not need to just destroy the object. It can implement any other operation like inserting the remove object in another container. Let's see a small example: [import ../example/doc_erasing_and_disposing.cpp] [doc_erasing_and_disposing] All [*Boost.Intrusive] containers offer these "erase + dispose" additional members for all functions that erase an element from the container. [endsect] [section:clone_from Cloning Boost.Intrusive containers] As previously mentioned, [*Boost.Intrusive] containers are [*non-copyable and non-assignable], because intrusive containers don't allocate memory at all. To implement a copy-constructor or assignment operator, the user must clone one by one all the elements of the container and insert them in another intrusive container. However, cloning by hand is usually more inefficient than a member cloning function and a specialized cloning function can offer more guarantees than the manual cloning (better exception safety guarantees, for example). To ease the implementation of copy constructors and assignment operators of classes containing [*Boost.Intrusive] containers, all [*Boost.Intrusive] containers offer a special cloning function called `clone_from`. Apart from the container to be cloned, `clone_from` takes two function objects as arguments. For example, consider the `clone_from` member function of [classref boost::intrusive::list list]: [c++] template void clone_from(const list &src, Cloner cloner, Disposer disposer); This function will make `*this` a clone of `src`. Let's explain the arguments: * The first parameter is the list to be cloned. * The second parameter is a function object that will clone `value_type` objects and return a pointer to the clone. It must implement the following function: `pointer operator()(const value_type &)`. * The second parameter is a function object that will dispose `value_type` objects. It's used first to empty the container before cloning and to dispose the elements if an exception is thrown. The cloning function works as follows: * First it clears and disposes all the elements from *this using the disposer function object. * After that it starts cloning all the elements of the source container using the cloner function object. * If any operation in the cloning function (for example, the cloner function object) throws, all the constructed elements are disposed using the disposer function object. Here is an example of `clone_from`: [import ../example/doc_clone_from.cpp] [doc_clone_from] [endsect] [section:function_hooks Using function hooks] A programmer might find that base or member hooks are not flexible enough in some situations. In some applications it would be optimal to put a hook deep inside a member of a class or just outside the class. [*Boost.Intrusive] has an easy option to allow such cases: [classref boost::intrusive::function_hook function_hook]. This option is similar to [classref boost::intrusive::member_hook member_hook] or [classref boost::intrusive::base_hook base_hook], but the programmer can specify a function object that tells the container how to obtain a hook from a value and vice versa. The programmer just needs to define the following function object: [c++] //This functor converts between value_type and a hook_type struct Functor { //Required types typedef /*impl-defined*/ hook_type; typedef /*impl-defined*/ hook_ptr; typedef /*impl-defined*/ const_hook_ptr; typedef /*impl-defined*/ value_type; typedef /*impl-defined*/ pointer; typedef /*impl-defined*/ const_pointer; //Required static functions static hook_ptr to_hook_ptr (value_type &value); static const_hook_ptr to_hook_ptr(const value_type &value); static pointer to_value_ptr(hook_ptr n); static const_pointer to_value_ptr(const_hook_ptr n); }; Converting from values to hooks is generally easy, since most hooks are in practice members or base classes of class data members. The inverse operation is a bit more complicated, but [*Boost.Intrusive] offers a bit of help with the function [funcref boost::intrusive::get_parent_from_member get_parent_from_member], which allows easy conversions from the address of a data member to the address of the parent holding that member. Let's see a little example of [classref boost::intrusive::function_hook function_hook]: [import ../example/doc_function_hooks.cpp] [doc_function_hooks] [endsect] [section:recursive Recursive Boost.Intrusive containers] [*Boost.Intrusive] containers can be used to define recursive structures very easily, allowing complex data structures with very low overhead. Let's see an example: [import ../example/doc_recursive.cpp] [doc_recursive] Recursive data structures using [*Boost.Intrusive] containers must avoid using hook deduction to avoid early type instantiation: [c++] //This leads to compilation error (Recursive is instantiated by //'list' to deduce hook properties (pointer type, tag, safe-mode...) class Recursive { //... list< Recursive > l; //... }; //Ok, programmer must specify the hook type to avoid early Recursive instantiation class Recursive { //... list< Recursive, base_hook > l; //... }; Member hooks are not suitable for recursive structures: [c++] class Recursive { private: Recursive(const Recursive&); Recursive & operator=(const Recursive&); public: list_member_hook<> memhook; list< Recursive, member_hook, &Recursive::memhook> > children; }; Specifying `&Recursive::memhook` (that is, the offset between memhook and Recursive) provokes an early instantiation of `Recursive`. To define recursive structures using member hooks, a programmer should use [classref ::boost::interprocess::function_hook function_hook]: [import ../example/doc_recursive_member.cpp] [doc_recursive_member] [endsect] [section:using_smart_pointers Using smart pointers with Boost.Intrusive containers] [*Boost.Intrusive] hooks can be configured to use other pointers than raw pointers. When a [*Boost.Intrusive] hook is configured with a smart pointer as an argument, this pointer configuration is passed to the containers. For example, if the following hook is configured with a smart pointer (for example, an offset pointer from [*Boost.Interprocess]): [import ../example/doc_offset_ptr.cpp] [doc_offset_ptr_0] Any intrusive list constructed using this hook will be ready for shared memory, because the intrusive list will also use offset pointers internally. For example, we can create an intrusive list in shared memory combining [*Boost.Interprocess] and [*Boost.Intrusive]: [doc_offset_ptr_1] [section:smart_pointers_requirements Requirements for smart pointers compatible with Boost.Intrusive] Not every smart pointer is compatible with [*Boost.Intrusive]: * It must be compatible with C++11 [@http://en.cppreference.com/w/cpp/memory/pointer_traits `std::pointer_traits`] requirements. [*Boost.Intrusive] uses its own [classref boost::intrusive::pointer_traits pointer_traits] class to implement those features in both C++11 and C++03 compilers. * It must have the same ownership semantics as a raw pointer. This means that resource management smart pointers (like `boost::shared_ptr`) can't be used. The conversion from the smart pointer to a raw pointer will be implemented as a recursive call to `operator->()` until the function returns a raw pointer. [endsect] [endsect] [section:obtaining_iterators_from_values Obtaining iterators from values] [*Boost.Intrusive] offers another useful feature that's not present in STL containers: it's possible to obtain an iterator to a value from the value itself. This feature is implemented in [*Boost.Intrusive] containers by a function called `iterator_to`: [c++] iterator iterator_to(reference value); const_iterator iterator_to(const_reference value); For [*Boost.Intrusive] containers that have local iterators, like unordered associative containers, we can also obtain local iterators: [c++] local_iterator local_iterator_to(reference value); const_local_iterator local_iterator_to(const_reference value) const; For most [*Boost.Intrusive] containers ([classref boost::intrusive::list list], [classref boost::intrusive::slist slist], [classref boost::intrusive::set set], [classref boost::intrusive::multiset multiset]) we have an alternative static `s_iterator_to` function. For unordered associative containers ([classref boost::intrusive::unordered_set unordered_set], [classref boost::intrusive::multiset multiset]), `iterator_to` has no static alternative function. On the other hand, `local_iterator_to` functions have their `s_local_iterator_to` static alternatives. Alternative static functions are available under certain circumstances explained in the [link intrusive.value_traits.stateful_value_traits Stateful value traits] section; if the programmer uses hooks provided by [*Boost.Intrusive], those functions will be available. Let's see a small function that shows the use of `iterator_to` and `local_iterator_to`: [import ../example/doc_iterator_from_value.cpp] [doc_iterator_from_value] [endsect] [section:any_hooks Any Hooks: A single hook for any Intrusive container] Sometimes, a class programmer wants to place a class in several intrusive containers but no at the same time. In this case, the programmer might decide to insert two hooks in the same class. [c++] class MyClass : public list_base_hook<>, public slist_base_hook<> //... {}; However, there is a more size-efficient alternative in [*Boost.Intrusive]: "any" hooks ([classref boost::intrusive::any_base_hook any_base_hook] and [classref boost::intrusive::any_member_hook any_member_hook]). These hooks can be used to store a type in several containers offered by [*Boost.Intrusive] minimizing the size of the class. These hooks support these options: * [*`tag`] (for base hooks only): This argument serves as a tag, so you can derive from more than one slist hook. Default: `tag`. * [*`link_mode`]: The linking policy. `link_mode` is [*not] supported and `link_mode` might offer weaker error detection in any hooks than in other hooks. Default: `link_mode`. * [*`void_pointer`]: The pointer type to be used internally in the hook and propagated to the container. Default: `void_pointer`. `auto_unlink` can't be supported because the hook does not know in which type of container might be currently inserted. Additionally, these hooks don't support `unlink()` and `swap_nodes()` operations for the same reason. Here is an example that creates a class with two any hooks, and uses one to insert the class in a [classref slist] and the other one in a [classref list]. [import ../example/doc_any_hook.cpp] [doc_any_hook] [endsect] [section:concepts Concepts explained] This section will expand the explanation of previously presented basic concepts before explaining the customization options of [*Boost.Intrusive]. * [*Node Algorithms]: A set of static functions that implement basic operations on a group of nodes: initialize a node, link a node to a group of nodes, unlink a node from another group of nodes, etc. For example, a circular singly linked list is a group of nodes, where each node has a pointer to the next node. [*Node Algorithms] just require a [*NodeTraits] template parameter and they can work with any [*NodeTraits] class that fulfills the needed interface. As an example, here is a class that implements operations to manage a group of nodes forming a circular singly linked list: [c++] template struct my_slist_algorithms { typedef typename NodeTraits::node_ptr node_ptr; typedef typename NodeTraits::const_node_ptr const_node_ptr; //Get the previous node of "this_node" static node_ptr get_prev_node(node_ptr this_node) { node_ptr p = this_node; while (this_node != NodeTraits::get_next(p)) p = NodeTraits::get_next(p); return p; } // number of elements in the group of nodes containing "this_node" static std::size_t count(const_node_ptr this_node) { std::size_t result = 0; const_node_ptr p = this_node; do{ p = NodeTraits::get_next(p); ++result; } while (p != this_node); return result; } // More operations // ... }; * [*Node Traits]: A class that encapsulates the basic information and operations on a node within a group of nodes: the type of the node, a function to obtain the pointer to the next node, etc. [*Node Traits] specify the configuration information [*Node Algorithms] need. Each type of [*Node Algorithm] expects an interface that compatible [*Node Traits] classes must implement. As an example, this is the definition of a [*Node Traits] class that is compatible with the previously presented `my_slist_algorithms`: [c++] struct my_slist_node_traits { //The type of the node struct node { node *next_; }; typedef node * node_ptr; typedef const node * const_node_ptr; //A function to obtain a pointer to the next node static node_ptr get_next(const_node_ptr n) { return n->next_; } //A function to set the pointer to the next node static void set_next(node_ptr n, node_ptr next) { n->next_ = next; } }; * [*Hook]: A class that the user must add as a base class or as a member to his own class to make that class insertable in an intrusive container. Usually the hook contains a node object that will be used to form the group of nodes: For example, the following class is a [*Hook] that the user can add as a base class, to make the user class compatible with a singly linked list container: [c++] class my_slist_base_hook //This hook contains a node, that will be used //to link the user object in the group of nodes : private my_slist_node_traits::node { typedef my_slist_node_traits::node_ptr node_ptr; typedef my_slist_node_traits::const_node_ptr const_node_ptr; //Converts the generic node to the hook static my_slist_base_hook *to_hook_ptr(node_ptr p) { return static_cast(p); } //Returns the generic node stored by this hook node_ptr to_node_ptr() { return static_cast(this); } // More operations // ... }; //To make MyClass compatible with an intrusive singly linked list //derive our class from the hook. class MyClass : public my_slist_base_hook { void set(int value); int get() const; private: int value_; }; * [*Intrusive Container]: A container that offers a STL-like interface to store user objects. An intrusive container can be templatized to store different value types that use different hooks. An intrusive container is also more elaborate than a group of nodes: it can store the number of elements to achieve constant-time size information, it can offer debugging facilities, etc. For example, an [classref boost::intrusive::slist slist] container (intrusive singly linked list) should be able to hold `MyClass` objects that might have decided to store the hook as a base class or as a member. Internally, the container will use [*Node Algorithms] to implement its operations, and an intrusive container is configured using a template parameter called [*ValueTraits]. [*ValueTraits] will contain the information to convert user classes in nodes compatible with [*Node Algorithms]. For example, this a possible [classref boost::intrusive::slist slist] implementation: [c++] template class slist { public: typedef typename ValueTraits::value_type value_type; //More typedefs and functions // ... //Insert the value as the first element of the list void push_front (reference value) { node_ptr to_insert(ValueTraits::to_node_ptr(value)); circular_list_algorithms::link_after(to_insert, get_root_node()); } // More operations // ... }; * [*Semi-Intrusive Container]: A semi-intrusive container is similar to an intrusive container, but apart from the values to be inserted in the container, it needs additional memory (for example, auxiliary arrays or indexes). * [*Value Traits]: As we can see, to make our classes intrusive-friendly we add a simple hook as a member or base class. The hook contains a generic node that will be inserted in a group of nodes. [*Node Algorithms] just work with nodes and don't know anything about user classes. On the other hand, an intrusive container needs to know how to obtain a node from a user class, and also the inverse operation. So we can define [*ValueTraits] as the glue between user classes and nodes required by [*Node Algorithms]. Let's see a possible implementation of a value traits class that glues MyClass and the node stored in the hook: [c++] struct my_slist_derivation_value_traits { public: typedef slist_node_traits node_traits; typedef MyClass value_type; typedef node_traits::node_ptr node_ptr; typedef value_type* pointer; //... //Converts user's value to a generic node static node_ptr to_node_ptr(reference value) { return static_cast(value).to_node_ptr(); } //Converts a generic node into user's value static value_type *to_value_ptr(node_traits::node *n) { static_cast(slist_base_hook::to_hook_ptr(n)); } // More operations // ... }; [endsect] [section:node_algorithms Node algorithms with custom NodeTraits] As explained in the [link intrusive.concepts Concepts] section, [*Boost.Intrusive] containers are implemented using node algorithms that work on generic nodes. Sometimes, the use of intrusive containers is expensive for some environments and the programmer might want to avoid all the template instantiations related to [*Boost.Intrusive] containers. However, the user can still benefit from [*Boost.Intrusive] using the node algorithms, because some of those algorithms, like red-black tree algorithms, are not trivial to write. All node algorithm classes are templatized by a `NodeTraits` class. This class encapsulates the needed internal type declarations and operations to make a node compatible with node algorithms. Each type of node algorithms has its own requirements: [section:circular_slist_algorithms Intrusive singly linked list algorithms] These algorithms are static members of the [classref boost::intrusive::circular_slist_algorithms circular_slist_algorithms] class: [c++] template struct circular_slist_algorithms; An empty list is formed by a node whose pointer to the next node points to itself. [classref boost::intrusive::circular_slist_algorithms circular_slist_algorithms] is configured with a NodeTraits class, which encapsulates the information about the node to be manipulated. NodeTraits must support the following interface: [*Typedefs]: * `node`: The type of the node that forms the circular list * `node_ptr`: The type of a pointer to a node (usually node*) * `const_node_ptr`: The type of a pointer to a const node (usually const node*) [*Static functions]: * `static node_ptr get_next(const_node_ptr n);`: Returns a pointer to the next node stored in "n". * `static void set_next(node_ptr n, node_ptr next);`: Sets the pointer to the next node stored in "n" to "next". Once we have a node traits configuration we can use [*Boost.Intrusive] algorithms with our nodes: [import ../example/doc_slist_algorithms.cpp] [doc_slist_algorithms_code] For a complete list of functions see [classref boost::intrusive::circular_slist_algorithms circular_slist_algorithms reference]. [endsect] [section:circular_list_algorithms Intrusive doubly linked list algorithms] These algorithms are static members of the [classref boost::intrusive::circular_list_algorithms circular_list_algorithms] class: [c++] template struct circular_list_algorithms; An empty list is formed by a node whose pointer to the next node points to itself. [classref boost::intrusive::circular_list_algorithms circular_list_algorithms] is configured with a NodeTraits class, which encapsulates the information about the node to be manipulated. NodeTraits must support the following interface: [*Typedefs]: * `node`: The type of the node that forms the circular list * `node_ptr`: The type of a pointer to a node (usually node*) * `const_node_ptr`: The type of a pointer to a const node (usually const node*) [*Static functions]: * `static node_ptr get_next(const_node_ptr n);`: Returns a pointer to the next node stored in "n". * `static void set_next(node_ptr n, node_ptr next);`: Sets the pointer to the next node stored in "n" to "next". * `static node_ptr get_previous(const_node_ptr n);`: Returns a pointer to the previous node stored in "n". * `static void set_previous(node_ptr n, node_ptr prev);`: Sets the pointer to the previous node stored in "n" to "prev". Once we have a node traits configuration we can use [*Boost.Intrusive] algorithms with our nodes: [import ../example/doc_list_algorithms.cpp] [doc_list_algorithms_code] For a complete list of functions see [classref boost::intrusive::circular_list_algorithms circular_list_algorithms reference]. [endsect] [section:rbtree_algorithms Intrusive red-black tree algorithms] These algorithms are static members of the [classref boost::intrusive::rbtree_algorithms rbtree_algorithms] class: [c++] template struct rbtree_algorithms; An empty tree is formed by a node whose pointer to the parent node is null, the left and right node pointers point to itself, and whose color is red. [classref boost::intrusive::rbtree_algorithms rbtree_algorithms] is configured with a NodeTraits class, which encapsulates the information about the node to be manipulated. NodeTraits must support the following interface: [*Typedefs]: * `node`: The type of the node that forms the circular rbtree * `node_ptr`: The type of a pointer to a node (usually node*) * `const_node_ptr`: The type of a pointer to a const node (usually const node*) * `color`: The type that can store the color of a node [*Static functions]: * `static node_ptr get_parent(const_node_ptr n);`: Returns a pointer to the parent node stored in "n". * `static void set_parent(node_ptr n, node_ptr p);`: Sets the pointer to the parent node stored in "n" to "p". * `static node_ptr get_left(const_node_ptr n);`: Returns a pointer to the left node stored in "n". * `static void set_left(node_ptr n, node_ptr l);`: Sets the pointer to the left node stored in "n" to "l". * `static node_ptr get_right(const_node_ptr n);`: Returns a pointer to the right node stored in "n". * `static void set_right(node_ptr n, node_ptr r);`: Sets the pointer to the right node stored in "n" to "r". * `static color get_color(const_node_ptr n);`: Returns the color stored in "n". * `static void set_color(node_ptr n, color c);`: Sets the color stored in "n" to "c". * `static color black();`: Returns a value representing the black color. * `static color red();`: Returns a value representing the red color. Once we have a node traits configuration we can use [*Boost.Intrusive] algorithms with our nodes: [import ../example/doc_rbtree_algorithms.cpp] [doc_rbtree_algorithms_code] For a complete list of functions see [classref boost::intrusive::rbtree_algorithms rbtree_algorithms reference]. [endsect] [section:splaytree_algorithms Intrusive splay tree algorithms] These algorithms are static members of the [classref boost::intrusive::splaytree_algorithms splaytree_algorithms] class: [c++] template struct splaytree_algorithms; An empty tree is formed by a node whose pointer to the parent node is null, and whose left and right nodes pointers point to itself. [classref boost::intrusive::splaytree_algorithms splaytree_algorithms] is configured with a NodeTraits class, which encapsulates the information about the node to be manipulated. NodeTraits must support the following interface: [*Typedefs]: * `node`: The type of the node that forms the circular splaytree * `node_ptr`: The type of a pointer to a node (usually node*) * `const_node_ptr`: The type of a pointer to a const node (usually const node*) [*Static functions]: * `static node_ptr get_parent(const_node_ptr n);`: Returns a pointer to the parent node stored in "n". * `static void set_parent(node_ptr n, node_ptr p);`: Sets the pointer to the parent node stored in "n" to "p". * `static node_ptr get_left(const_node_ptr n);`: Returns a pointer to the left node stored in "n". * `static void set_left(node_ptr n, node_ptr l);`: Sets the pointer to the left node stored in "n" to "l". * `static node_ptr get_right(const_node_ptr n);`: Returns a pointer to the right node stored in "n". * `static void set_right(node_ptr n, node_ptr r);`: Sets the pointer to the right node stored in "n" to "r". Once we have a node traits configuration we can use [*Boost.Intrusive] algorithms with our nodes: [import ../example/doc_splaytree_algorithms.cpp] [doc_splaytree_algorithms_code] For a complete list of functions see [classref boost::intrusive::splaytree_algorithms splaytree_algorithms reference]. [endsect] [section:avltree_algorithms Intrusive avl tree algorithms] [classref boost::intrusive::avltree_algorithms avltree_algorithms] have the same interface as [classref boost::intrusive::rbtree_algorithms rbtree_algorithms]. [c++] template struct avltree_algorithms; [classref boost::intrusive::avltree_algorithms avltree_algorithms] is configured with a NodeTraits class, which encapsulates the information about the node to be manipulated. NodeTraits must support the following interface: [*Typedefs]: * `node`: The type of the node that forms the circular avltree * `node_ptr`: The type of a pointer to a node (usually node*) * `const_node_ptr`: The type of a pointer to a const node (usually const node*) * `balance`: A type that can represent 3 balance types (usually an integer) [*Static functions]: * `static node_ptr get_parent(const_node_ptr n);`: Returns a pointer to the parent node stored in "n". * `static void set_parent(node_ptr n, node_ptr p);`: Sets the pointer to the parent node stored in "n" to "p". * `static node_ptr get_left(const_node_ptr n);`: Returns a pointer to the left node stored in "n". * `static void set_left(node_ptr n, node_ptr l);`: Sets the pointer to the left node stored in "n" to "l". * `static node_ptr get_right(const_node_ptr n);`: Returns a pointer to the right node stored in "n". * `static void set_right(node_ptr n, node_ptr r);`: Sets the pointer to the right node stored in "n" to "r". * `static balance get_balance(const_node_ptr n);`: Returns the balance factor stored in "n". * `static void set_balance(node_ptr n, balance b);`: Sets the balance factor stored in "n" to "b". * `static balance negative();`: Returns a value representing a negative balance factor. * `static balance zero();`: Returns a value representing a zero balance factor. * `static balance positive();`: Returns a value representing a positive balance factor. Once we have a node traits configuration we can use [*Boost.Intrusive] algorithms with our nodes: [import ../example/doc_avltree_algorithms.cpp] [doc_avltree_algorithms_code] For a complete list of functions see [classref boost::intrusive::avltree_algorithms avltree_algorithms reference]. [endsect] [section:treap_algorithms Intrusive treap algorithms] [classref boost::intrusive::treap_algorithms treap_algorithms] have the same interface as [classref boost::intrusive::rbtree_algorithms rbtree_algorithms]. [c++] template struct treap_algorithms; [classref boost::intrusive::treap_algorithms treap_algorithms] is configured with a NodeTraits class, which encapsulates the information about the node to be manipulated. NodeTraits must support the following interface: [*Typedefs]: * `node`: The type of the node that forms the circular treap * `node_ptr`: The type of a pointer to a node (usually node*) * `const_node_ptr`: The type of a pointer to a const node (usually const node*) [*Static functions]: * `static node_ptr get_parent(const_node_ptr n);`: Returns a pointer to the parent node stored in "n". * `static void set_parent(node_ptr n, node_ptr p);`: Sets the pointer to the parent node stored in "n" to "p". * `static node_ptr get_left(const_node_ptr n);`: Returns a pointer to the left node stored in "n". * `static void set_left(node_ptr n, node_ptr l);`: Sets the pointer to the left node stored in "n" to "l". * `static node_ptr get_right(const_node_ptr n);`: Returns a pointer to the right node stored in "n". * `static void set_right(node_ptr n, node_ptr r);`: Sets the pointer to the right node stored in "n" to "r". Once we have a node traits configuration we can use [*Boost.Intrusive] algorithms with our nodes: [import ../example/doc_treap_algorithms.cpp] [doc_treap_algorithms_code] For a complete list of functions see [classref boost::intrusive::treap_algorithms treap_algorithms reference]. [endsect] [/ / /[section:sgtree_algorithms Intrusive sg tree algorithms] / / /[classref boost::intrusive::sgtree_algorithms sgtree_algorithms] have the same /interface as [classref boost::intrusive::rbtree_algorithms rbtree_algorithms]. / /[c++] / / template / struct sgtree_algorithms; / /[classref boost::intrusive::sgtree_algorithms sgtree_algorithms] /is configured with a NodeTraits class, which encapsulates /the information about the node to be manipulated. NodeTraits must support the /following interface: / /[*Typedefs]: / /* `node`: The type of the node that forms the circular sgtree / /* `node_ptr`: The type of a pointer to a node (usually node*) / /* `const_node_ptr`: The type of a pointer to a const node (usually const node*) / /[*Static functions]: / /* `static node_ptr get_parent(const_node_ptr n);`: / Returns a pointer to the parent node stored in "n". / /* `static void set_parent(node_ptr n, node_ptr p);`: / Sets the pointer to the parent node stored in "n" to "p". / /* `static node_ptr get_left(const_node_ptr n);`: / Returns a pointer to the left node stored in "n". / /* `static void set_left(node_ptr n, node_ptr l);`: / Sets the pointer to the left node stored in "n" to "l". / /* `static node_ptr get_right(const_node_ptr n);`: / Returns a pointer to the right node stored in "n". / /* `static void set_right(node_ptr n, node_ptr r);`: / Sets the pointer to the right node stored in "n" to "r". / /Once we have a node traits configuration we can use [*Boost.Intrusive] algorithms /with our nodes: / /[import ../example/doc_sgtree_algorithms.cpp] /[doc_sgtree_algorithms_code] / /For a complete list of functions see /[classref boost::intrusive::sgtree_algorithms sgtree_algorithms reference]. / /[endsect] /] [endsect] [section:value_traits Containers with custom ValueTraits] As explained in the [link intrusive.concepts Concepts] section, [*Boost.Intrusive] containers need a `ValueTraits` class to perform transformations between nodes and user values. `ValueTraits` can be explicitly configured (using the `value_traits<>` option) or implicitly configured (using hooks and their `base_hook<>`/`member_hook<>` options). `ValueTraits` contains all the information to glue the `value_type` of the containers and the node to be used in node algorithms, since these types can be different. Apart from this, `ValueTraits` also stores information about the link policy of the values to be inserted. Instead of using [*Boost.Intrusive] predefined hooks a user might want to develop customized containers, for example, using nodes that are optimized for a specific application or that are compatible with a legacy ABI. A user might want to have only two additional pointers in his class and insert the class in a doubly linked list sometimes and in a singly linked list in other situations. You can't achieve this using [*Boost.Intrusive] predefined hooks. Now, instead of using `base_hook<...>` or `member_hook<...>` options the user will specify the `value_traits<...>` options. Let's see how we can do this: [section:value_traits_interface ValueTraits interface] `ValueTraits` has the following interface: [c++] #include #include struct my_value_traits { typedef implementation_defined node_traits; typedef implementation_defined value_type; typedef node_traits::node_ptr node_ptr; typedef node_traits::const_node_ptr const_node_ptr; typedef boost::intrusive::pointer_traits::rebind_traits ::type::pointer pointer; typedef boost::intrusive::pointer_traits::rebind_traits ::type::pointer const_pointer; static const link_mode_type link_mode = some_linking_policy; static node_ptr to_node_ptr (value_type &value); static const_node_ptr to_node_ptr (const value_type &value); static pointer to_value_ptr (node_ptr n); static const_pointer to_value_ptr (const_node_ptr n); }; Let's explain each type and function: * [*['node_traits]]: The node configuration that is needed by node algorithms. These node traits and algorithms are described in the previous chapter: [link intrusive.node_algorithms Node Algorithms]. * If my_value_traits is meant to be used with [classref boost::intrusive::slist slist], `node_traits` should follow the interface needed by [classref boost::intrusive::circular_slist_algorithms circular_slist_algorithms]. * If my_value_traits is meant to be used with [classref boost::intrusive::list list], `node_traits` should follow the interface needed by [classref boost::intrusive::circular_list_algorithms circular_list_algorithms]. * If my_value_traits is meant to be used with [classref boost::intrusive::set set]/[classref boost::intrusive::multiset multiset], `node_traits` should follow the interface needed by [classref boost::intrusive::rbtree_algorithms rbtree_algorithms]. * If my_value_traits is meant to be used with [classref boost::intrusive::unordered_set unordered_set]/ [classref boost::intrusive::unordered_multiset unordered_multiset], `node_traits` should follow the interface needed by [classref boost::intrusive::circular_slist_algorithms circular_slist_algorithms]. * [*['node_ptr]]: A typedef for `node_traits::node_ptr`. * [*['const_node_ptr]]: A typedef for `node_traits::const_node_ptr`. * [*['value_type]]: The type that the user wants to insert in the container. This type can be the same as `node_traits::node` but it can be different (for example, `node_traits::node` can be a member type of `value_type`). If `value_type` and `node_traits::node` are the same type, the `to_node_ptr` and `to_value_ptr` functions are trivial. * [*['pointer]]: The type of a pointer to a `value_type`. It must be the same pointer type as `node_ptr`: If `node_ptr` is `node*`, `pointer` must be `value_type*`. If `node_ptr` is `smart_ptr`, `pointer` must be `smart_ptr`. This can be generically achieved using `boost::intrusive::pointer_traits` (portable implementation of C++11 `std::pointer_traits`). * [*['const_pointer]]: The type of a pointer to a `const value_type`. It must be the same pointer type as `node_ptr`: If `node_ptr` is `node*`, `const_pointer` must be `const value_type*`. If `node_ptr` is `smart_ptr`, `const_pointer` must be `smart_ptr`. * [*['link_mode]]: Indicates that `value_traits` needs some additional work or checks from the container. The types are enumerations defined in the `link_mode.hpp` header. These are the possible types: * [*`normal_link`]: If this linking policy is specified in a `ValueTraits` class as the link mode, containers configured with such `ValueTraits` won't set the hooks of the erased values to a default state. Containers also won't check that the hooks of the new values are default initialized. * [*`safe_link`]: If this linking policy is specified as the link mode in a `ValueTraits` class, containers configured with this `ValueTraits` will set the hooks of the erased values to a default state. Containers also will check that the hooks of the new values are default initialized. * [*`auto_unlink`]: Same as "safe_link" but containers with constant-time size features won't be compatible with `ValueTraits` configured with this policy. Containers also know that a value can be silently erased from the container without using any function provided by the containers. * [*['static node_ptr to_node_ptr (value_type &value)]] and [*['static const_node_ptr to_node_ptr (const value_type &value)]]: These functions take a reference to a value_type and return a pointer to the node to be used with node algorithms. * [*['static pointer to_value_ptr (node_ptr n)]] and [*['static const_pointer to_value_ptr (const_node_ptr n)]]: These functions take a pointer to a node and return a pointer to the value that contains the node. [endsect] [section:value_traits_example Custom ValueTraits example] Let's define our own `value_traits` class to be able to use [*Boost.Intrusive] containers with an old C structure whose definition can't be changed. That legacy type has two pointers that can be used to build singly and doubly linked lists: in singly linked lists we only need a pointer, whereas in doubly linked lists, we need two pointers. Since we only have two pointers, we can't insert the object in both a singly and a doubly linked list at the same time. This is the definition of the old node: [import ../example/doc_value_traits.cpp] [doc_value_traits_code_legacy] Now we have to define a NodeTraits class that will implement the functions/typedefs that will make the legacy node compatible with [*Boost.Intrusive] algorithms. After that, we'll define a ValueTraits class that will configure [*Boost.Intrusive] containers: [doc_value_traits_value_traits] Defining a value traits class that simply defines `value_type` as `legacy_node_traits::node` is a common approach when defining customized intrusive containers, so [*Boost.Intrusive] offers a templatized [classref boost::intrusive::trivial_value_traits trivial_value_traits] class that does exactly what we want: [doc_value_traits_trivial] Now we can just define the containers that will store the legacy abi objects and write a little test: [doc_value_traits_test] As seen, several key elements of [*Boost.Intrusive] can be reused with custom user types, if the user does not want to use the provided [*Boost.Intrusive] facilities. [endsect] [section:reusing_node_algorithms Reusing node algorithms for different values] In the previous example, `legacy_node_traits::node` type and `legacy_value_traits::value_type` are the same type, but this is not necessary. It's possible to have several `ValueTraits` defining the same `node_traits` type (and thus, the same `node_traits::node`). This reduces the number of node algorithm instantiations, but now `ValueTraits::to_node_ptr` and `ValueTraits::to_value_ptr` functions need to offer conversions between both types. Let's see a small example: First, we'll define the node to be used in the algorithms. For a linked list, we just need a node that stores two pointers: [import ../example/doc_advanced_value_traits.cpp] [doc_advanced_value_traits_code] Now we'll define two different types that will be inserted in intrusive lists and a templatized `ValueTraits` that will work for both types: [doc_advanced_value_traits_value_traits] Now define two containers. Both containers will instantiate the same list algorithms (`circular_list_algorithms`), due to the fact that the value traits used to define the containers provide the same `node_traits` type: [doc_advanced_value_traits_containers] All [*Boost.Intrusive] containers using predefined hooks use this technique to minimize code size: all possible [classref boost::intrusive::list list] containers created with predefined hooks that define the same `VoidPointer` type share the same list algorithms. [endsect] [section:simplifying_value_traits Simplifying value traits definition] The previous example can be further simplified using the [classref boost::intrusive::derivation_value_traits derivation_value_traits] class to define a value traits class with a value that stores the `simple_node` as a base class: [import ../example/doc_derivation_value_traits.cpp] [doc_derivation_value_traits_value_traits] We can even choose to store `simple_node` as a member of `value_1` and `value_2` classes and use [classref boost::intrusive::member_value_traits member_value_traits] to define the needed value traits classes: [import ../example/doc_member_value_traits.cpp] [doc_member_value_traits_value_traits] [endsect] [section:stateful_value_traits Stateful value traits] Until now all shown custom value traits are stateless, that is, [*the transformation between nodes and values is implemented in terms of static functions]. It's possible to use [*stateful] value traits so that we can separate nodes and values and [*avoid modifying types to insert nodes]. [*Boost.Intrusive] differentiates between stateful and stateless value traits by checking if all Node <-> Value transformation functions are static or not (except for Visual 7.1, since overloaded static function detection is not possible, in this case the implementation checks if the class is empty): * If all Node <-> Value transformation functions are static , a [*stateless] value traits is assumed. transformations must be static functions. * Otherwise a [*stateful] value traits is assumed. Using stateful value traits it's possible to create containers of non-copyable/movable objects [*without modifying] the definition of the class to be inserted. This interesting property is achieved without using global variables (stateless value traits could use global variables to achieve the same goal), so: * [*Thread-safety guarantees]: Better thread-safety guarantees can be achieved with stateful value traits, since accessing global resources might require synchronization primitives that can be avoided when using internal state. * [*Flexibility]: A stateful value traits type can be configured at run-time. * [*Run-time polymorphism]: A value traits might implement node <-> value transformations as virtual functions. A single container type could be configured at run-time to use different node <-> value relationships. Stateful value traits have many advantages but also some downsides: * [*Performance]: Value traits operations should be very efficient since they are basic operations used by containers. [*A heavy node <-> value transformation will hurt intrusive containers' performance]. * [*Exception guarantees]: The stateful ValueTraits must maintain no-throw guarantees, otherwise, the container invariants won't be preserved. * [*Static functions]: Some static functions offered by intrusive containers are not available because node <-> value transformations are not static. * [*Bigger iterators]: The size of some iterators is increased because the iterator needs to store a pointer to the stateful value traits to implement node to value transformations (e.g. `operator*()` and `operator->()`). An easy and useful example of stateful value traits is when an array of values can be indirectly introduced in a list guaranteeing no additional allocation apart from the initial resource reservation: [import ../example/doc_stateful_value_traits.cpp] [doc_stateful_value_traits] [endsect] [endsect] [section:thread_safety Thread safety guarantees] Intrusive containers have thread safety guarantees similar to STL containers. * Several threads having read or write access to different instances is safe as long as inserted objects are different. * Concurrent read-only access to the same container is safe. Some Intrusive hooks (auto-unlink hooks, for example) modify containers without having a reference to them: this is considered a write access to the container. Other functions, like checking if an object is already inserted in a container using the `is_linked()` member of safe hooks, constitute read access on the container without having a reference to it, so no other thread should have write access (direct or indirect) to that container. Since the same object can be inserted in several containers at the same time using different hooks, the thread safety of [*Boost.Intrusive] is related to the containers and also to the object whose lifetime is manually managed by the user. As we can see, the analysis of the thread-safety of a program using [*Boost.Intrusive] is harder than with non-intrusive containers. To analyze the thread safety, consider the following points: * The auto-unlink hook's destructor and `unlink()` functions modify the container indirectly. * The safe mode and auto-unlink hooks' `is_linked()` functions are a read access to the container. * Inserting an object in containers that will be modified by different threads has no thread safety guarantee, although in most platforms it will be thread-safe without locking. [endsect] [section:boost_intrusive_iterators Boost.Intrusive Iterator features] [section:null_forward_iterators Null forward iterators] [*Boost.Intrusive] implements [@http://www.open-std.org/JTC1/sc22/WG21/docs/papers/2013/n3644.pdf C++14 Null Forward Iterators], a feature that was introduced with C++14. This means that equality and inequality comparison are defined over all iterators for the same underlying sequence and the value initialized-iterators. Value initialized iterators behave as if they refer past the end of the same empty sequence: [c++] list l = { ... }; auto ni = list::iterator(); auto nd = list::iterator(); ni == ni; // True. nd != nd; // False. ni == nd; // Won't compile. [endsect] [section:scary_iterators Scary Iterators] The paper N2913, titled [@http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2009/n2913.pdf, SCARY Iterator Assignment and Initialization], proposed a requirement that a standard container's iterator types have no dependency on any type argument apart from the container's `value_type`, `difference_type`, `pointer type`, and `const_pointer` type. In particular, according to the proposal, the types of a standard container's iterators should not depend on the container's `key_compare`, `hasher`, `key_equal`, or `allocator` types. That paper demonstrated that SCARY operations were crucial to the performant implementation of common design patterns using STL components. It showed that implementations that support SCARY operations reduce object code bloat by eliminating redundant specializations of iterator and algorithm templates. [*Boost.Intrusive] containers are a bit different from standard containers. In particular, they have no allocator parameter and they can be configured with additional options not present in STL-like containers. Thus [*Boost.Intrusive] offers its own `SCARY iterator` implementation, where iterator types don't change when the container is configured with an option that does not alter the value <-> node transformation. More concretely, the following options and conditions guarantee that iterator types are unchanged: * [*All containers]: `size_type<>`, `constant_time_size<>`, * [*`slist`]: `cache_last<>`, `linear<>`, * [*`unordered_[multi]set`]: `hash<>`, `equal<>`, `power_2_buckets<>`, `cache_begin<>`. * [*All tree-like containers] (`[multi]set`, `avl_[multi]set`, `sg_[multi]set`, `bs_[multi]set`, `splay_[multi]set`, `treap_[multi]set`): `compare<>`. * [*`treap_[multi]set`]: `priority<>` * [*`bs_[multi]set`, `sg_[multi]set`, `treap_[multi]set`, `splay_[multi]set`]: They share the same iterator type when configured with the same options. [endsect] [endsect] [section:equal_range_stability Stability and insertion with hint in ordered associative containers with equivalent keys] [*Boost.Intrusive] ordered associative containers with equivalent keys offer stability guarantees, following [@http://open-std.org/jtc1/sc22/wg21/docs/lwg-defects.html#233 C++ standard library's defect #233 resolution], explained in document [@http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2005/n1780.html Comments on LWG issue 233: Insertion hints in associative containers]. This means that: * A ['Insert without hint] member function always insert at the upper bound of an equal range. * A ['Insert with hint] member function inserts the new value [*before the hint] if hint's and new node's keys are equivalent. * Implements Andrew Koenig ['as close as possible to hint] proposal. A new element is always be inserted as close to the hint as possible. So, for example, if there is a subsequence of equivalent values, `a.begin()` as the hint means that the new element should be inserted before the subsequence even if `a.begin()` is far away. This allows code to always append (or prepend) an equal range with something as simple as: `m.insert(m.end(), new_node);` or `m.insert(m.begin(), new_node);` [endsect] [section:obtaining_same_type_reducing_space Obtaining the same types and reducing symbol length] The flexible option specification mechanism used by [*Boost.Intrusive] for hooks and containers has a couple of downsides: * If a user specifies the same options in different order or specifies some options and leaves the rest as defaults, the type of the created container/hook will be different. Sometimes this is annoying, because two programmers specifying the same options might end up with incompatible types. For example, the following two lists, although using the same options, do not have the same type: [c++] #include using namespace boost::intrusive; //Explicitly specify constant-time size and size type typedef list, size_type List1; //Implicitly specify constant-time size and size type typedef list List2; * Option specifiers lead to long template symbols for classes and functions. Option specifiers themselves are verbose and without variadic templates, several default template parameters are assigned for non-specified options. Object and debugging information files can grow and compilation times may suffer if long names are produced. To solve these issues [*Boost.Intrusive] offers some helper metafunctions that reduce symbol lengths and create the same type if the same options (either explicitly or implicitly) are used. These also improve compilation times. All containers and hooks have their respective `make_xxx` versions. The previously shown example can be rewritten like this to obtain the same list type: [c++] #include using namespace boost::intrusive; #include using namespace boost::intrusive; //Explicitly specify constant-time size and size type typedef make_list, size_type::type List1; //Implicitly specify constant-time size and size type typedef make_list::type List2; Produced symbol lengths and compilation times will usually be shorter and object/debug files smaller. If you are concerned with file sizes and compilation times, this option is your best choice. [endsect] [section:design_notes Design Notes] When designing [*Boost.Intrusive] the following guidelines have been taken into account: [section:performance_sensitive Boost.Intrusive in performance sensitive environments] [*Boost.Intrusive] should be a valuable tool in performance sensitive environments, and following this guideline, [*Boost.Intrusive] has been designed to offer well known complexity guarantees. Apart from that, some options, like optional constant-time, have been designed to offer faster complexity guarantees in some functions, like `slist::splice`. The advanced lookup and insertion functions for associative containers, taking an arbitrary key type and predicates, were designed to avoid unnecessary object constructions. [endsect] [section:space_constrained Boost.Intrusive in space constrained environments] [*Boost.Intrusive] should be useful in space constrained environments, and following this guideline [*Boost.Intrusive] separates node algorithms and intrusive containers to avoid instantiating node algorithms for each user type. For example, a single class of red-black algorithms will be instantiated to implement all set and multiset containers using raw pointers. This way, [*Boost.Intrusive] seeks to avoid any code size overhead associated with templates. Apart from that, [*Boost.Intrusive] implements some size improvements: for example, red-black trees embed the color bit in the parent pointer lower bit, if nodes are two-byte aligned. The option to forgo constant-time size operations can reduce container size, and this extra size optimization is noticeable when the container is empty or contains few values. [endsect] [section:basic_building_block Boost.Intrusive as a basic building block] [*Boost.Intrusive] can be a basic building block to build more complex containers and this potential has motivated many design decisions. For example, the ability to have more than one hook per user type opens the opportunity to implement multi-index containers on top of [*Boost.Intrusive]. [*Boost.Intrusive] containers implement advanced functions taking function objects as arguments (`clone_from`, `erase_and_dispose`, `insert_check`, etc.). These functions come in handy when implementing non-intrusive containers (for example, STL-like containers) on top of intrusive containers. [endsect] [section:extending_intrusive Extending Boost.Intrusive] [*Boost.Intrusive] offers a wide range of containers but also allows the construction of custom containers reusing [*Boost.Intrusive] elements. The programmer might want to use node algorithms directly or build special hooks that take advantage of an application environment. For example, the programmer can customize parts of [*Boost.Intrusive] to manage old data structures whose definition can't be changed. [endsect] [endsect] [section:performance Performance] [*Boost.Intrusive] containers offer speed improvements compared to non-intrusive containers primarily because: * They minimize memory allocation/deallocation calls. * They obtain better memory locality. This section will show performance tests comparing some operations on `boost::intrusive::list` and `std::list`: * Insertions using `push_back` and container destruction will show the overhead associated with memory allocation/deallocation. * The `reverse` member function will show the advantages of the compact memory representation that can be achieved with intrusive containers. * The `sort` and `write access` tests will show the advantage of intrusive containers minimizing memory accesses compared to containers of pointers. Given an object of type `T`, [classref boost::intrusive::list boost::intrusive::list] can replace `std::list` to avoid memory allocation overhead, or it can replace `std::list` when the user wants containers with polymorphic values or wants to share values between several containers. Because of this versatility, the performance tests will be executed for 6 different list types: * 3 intrusive lists holding a class named `itest_class`, each one with a different linking policy (`normal_link`, `safe_link`, `auto_unlink`). The `itest_class` objects will be tightly packed in a `std::vector` object. * `std::list`, where `test_class` is exactly the same as `itest_class`, but without the intrusive hook. * `std::list`. The list will contain pointers to `test_class` objects tightly packed in a `std::vector` object. We'll call this configuration ['compact pointer list] * `std::list`. The list will contain pointers to `test_class` objects owned by a `std::list` object. We'll call this configuration ['disperse pointer list]. Both `test_class` and `itest_class` are templatized classes that can be configured with a boolean to increase the size of the object. This way, the tests can be executed with small and big objects. Here is the first part of the testing code, which shows the definitions of `test_class` and `itest_class` classes, and some other utilities: [import ../perf/perf_list.cpp] [perf_list_value_type] As we can see, `test_class` is a very simple class holding an `int`. `itest_class` is just a class that has a base hook ([classref boost::intrusive::list_base_hook list_base_hook]) and also derives from `test_class`. `func_ptr_adaptor` is just a functor adaptor to convert function objects taking `test_list` objects to function objects taking pointers to them. You can find the full test code in the [@../../libs/intrusive/perf/perf_list.cpp perf_list.cpp] source file. [section:performance_results_push_back Back insertion and destruction] The first test will measure the benefits we can obtain with intrusive containers avoiding memory allocations and deallocations. All the objects to be inserted in intrusive containers are allocated in a single allocation call, whereas `std::list` will need to allocate memory for each object and deallocate it for every erasure (or container destruction). Let's compare the code to be executed for each container type for different insertion tests: [perf_list_push_back_intrusive] For intrusive containers, all the values are created in a vector and after that inserted in the intrusive list. [perf_list_push_back_stdlist] For a standard list, elements are pushed back using push_back(). [perf_list_push_back_stdptrlist] For a standard compact pointer list, elements are created in a vector and pushed back in the pointer list using push_back(). [perf_list_push_back_disperse_stdptrlist] For a ['disperse pointer list], elements are created in a list and pushed back in the pointer list using push_back(). These are the times in microseconds for each case, and the normalized time: [table Back insertion + destruction times for Visual C++ 7.1 / Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [5000 / 22500] [1 / 1]] [[`safe_link` intrusive list] [7812 / 32187] [1.56 / 1.43]] [[`auto_unlink` intrusive list] [10156 / 41562] [2.03 / 1.84]] [[Standard list] [26875 / 97500] [5.37 / 4.33]] [[Standard compact pointer list] [76406 / 86718] [15.28 / 3.85]] [[Standard disperse pointer list] [146562 / 175625] [29.31 / 7.80]] ] [table Back insertion + destruction times for GCC 4.1.1 / MinGW over Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [4375 / 22187] [1 / 1]] [[`safe_link` intrusive list] [7812 / 32812] [1.78 / 1.47]] [[`auto_unlink` intrusive list] [10468 / 42031] [2.39 / 1.89]] [[Standard list] [81250 / 98125] [18.57 / 4.42]] [[Standard compact pointer list] [83750 / 94218] [19.14 / 4.24]] [[Standard disperse pointer list] [155625 / 175625] [35.57 / 7.91]] ] [table Back insertion + destruction times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2) [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [4792 / 20495] [1 / 1]] [[`safe_link` intrusive list] [7709 / 30803] [1.60 / 1.5]] [[`auto_unlink` intrusive list] [10180 / 41183] [2.12 / 2.0]] [[Standard list] [17031 / 32586] [3.55 / 1.58]] [[Standard compact pointer list] [27221 / 34823] [5.68 / 1.69]] [[Standard disperse pointer list] [102272 / 60056] [21.34 / 2.93]] ] The results are logical: intrusive lists just need one allocation. The destruction time of the `normal_link` intrusive container is trivial (complexity: `O(1)`), whereas `safe_link` and `auto_unlink` intrusive containers need to put the hooks of erased values in the default state (complexity: `O(NumElements)`). That's why `normal_link` intrusive list shines in this test. Non-intrusive containers need to make many more allocations and that's why they lag behind. The `disperse pointer list` needs to make `NumElements*2` allocations, so the result is not surprising. The Linux test shows that standard containers perform very well against intrusive containers with big objects. Nearly the same GCC version in MinGW performs worse, so maybe a good memory allocator is the reason for these excellent results. [endsect] [section:performance_results_reversing Reversing] The next test measures the time needed to complete calls to the member function `reverse()`. Values (`test_class` and `itest_class`) and lists are created as explained in the previous section. Note that for pointer lists, `reverse` [*does not need to access `test_class` values stored in another list or vector], since this function just needs to adjust internal pointers, so in theory all tested lists need to perform the same operations. These are the results: [table Reverse times for Visual C++ 7.1 / Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [2656 / 10625] [1 / 1.83]] [[`safe_link` intrusive list] [2812 / 10937] [1.05 / 1.89]] [[`auto_unlink` intrusive list] [2710 / 10781] [1.02 / 1.86]] [[Standard list] [5781 / 14531] [2.17 / 2.51]] [[Standard compact pointer list] [5781 / 5781] [2.17 / 1]] [[Standard disperse pointer list] [10781 / 15781] [4.05 / 2.72]] ] [table Reverse times for GCC 4.1.1 / MinGW over Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [2656 / 10781] [1 / 2.22]] [[`safe_link` intrusive list] [2656 / 10781] [1 / 2.22]] [[`auto_unlink` intrusive list] [2812 / 10781] [1.02 / 2.22]] [[Standard list] [4843 / 12500] [1.82 / 2.58]] [[Standard compact pointer list] [4843 / 4843] [1.82 / 1]] [[Standard disperse pointer list] [9218 / 12968] [3.47 / 2.67]] ] [table Reverse times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2) [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [2742 / 10847] [1 / 3.41]] [[`safe_link` intrusive list] [2742 / 10847] [1 / 3.41]] [[`auto_unlink` intrusive list] [2742 / 11027] [1 / 3.47]] [[Standard list] [3184 / 10942] [1.16 / 3.44]] [[Standard compact pointer list] [3207 / 3176] [1.16 / 1]] [[Standard disperse pointer list] [5814 / 13381] [2.12 / 4.21]] ] For small objects the results show that the compact storage of values in intrusive containers improve locality and reversing is faster than with standard containers, whose values might be dispersed in memory because each value is independently allocated. Note that the dispersed pointer list (a list of pointers to values allocated in another list) suffers more because nodes of the pointer list might be more dispersed, since allocations from both lists are interleaved in the code: [c++] //Object list (holding `test_class`) stdlist objects; //Pointer list (holding `test_class` pointers) stdptrlist l; for(int i = 0; i < NumElements; ++i){ //Allocation from the object list objects.push_back(stdlist::value_type(i)); //Allocation from the pointer list l.push_back(&objects.back()); } For big objects the compact pointer list wins because the reversal test doesn't need access to values stored in another container. Since all the allocations for nodes of this pointer list are likely to be close (since there is no other allocation in the process until the pointer list is created) locality is better than with intrusive containers. The dispersed pointer list, as with small values, has poor locality. [endsect] [section:performance_results_sorting Sorting] The next test measures the time needed to complete calls to the member function `sort(Pred pred)`. Values (`test_class` and `itest_class`) and lists are created as explained in the first section. The values will be sorted in ascending and descending order each iteration. For example, if ['l] is a list: [c++] for(int i = 0; i < NumIter; ++i){ if(!(i % 2)) l.sort(std::greater()); else l.sort(std::less()); } For a pointer list, the function object will be adapted using `func_ptr_adaptor`: [c++] for(int i = 0; i < NumIter; ++i){ if(!(i % 2)) l.sort(func_ptr_adaptor >()); else l.sort(func_ptr_adaptor >()); } Note that for pointer lists, `sort` will take a function object that [*will access `test_class` values stored in another list or vector], so pointer lists will suffer an extra indirection: they will need to access the `test_class` values stored in another container to compare two elements. These are the results: [table Sort times for Visual C++ 7.1 / Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [16093 / 38906] [1 / 1]] [[`safe_link` intrusive list] [16093 / 39062] [1 / 1]] [[`auto_unlink` intrusive list] [16093 / 38906] [1 / 1]] [[Standard list] [32343 / 56406] [2.0 / 1.44]] [[Standard compact pointer list] [33593 / 46093] [2.08 / 1.18]] [[Standard disperse pointer list] [46875 / 68593] [2.91 / 1.76]] ] [table Sort times for GCC 4.1.1 / MinGW over Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [15000 / 39218] [1 / 1]] [[`safe_link` intrusive list] [15156 / 39531] [1.01 / 1.01]] [[`auto_unlink` intrusive list] [15156 / 39531] [1.01 / 1.01]] [[Standard list] [34218 / 56875] [2.28 / 1.45]] [[Standard compact pointer list] [35468 / 49218] [2.36 / 1.25]] [[Standard disperse pointer list] [47656 / 70156] [3.17 / 1.78]] ] [table Sort times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2) [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [18003 / 40795] [1 / 1]] [[`safe_link` intrusive list] [18003 / 41017] [1 / 1]] [[`auto_unlink` intrusive list] [18230 / 40941] [1.01 / 1]] [[Standard list] [26273 / 49643] [1.45 / 1.21]] [[Standard compact pointer list] [28540 / 43172] [1.58 / 1.05]] [[Standard disperse pointer list] [35077 / 57638] [1.94 / 1.41]] ] The results show that intrusive containers are faster than standard containers. We can see that the pointer list holding pointers to values stored in a vector is quite fast, so the extra indirection that is needed to access the value is minimized because all the values are tightly stored, improving caching. The disperse list, on the other hand, is slower because the indirection to access values stored in the object list is more expensive than accessing values stored in a vector. [endsect] [section:performance_results_write_access Write access] The next test measures the time needed to iterate through all the elements of a list, and increment the value of the internal `i_` member: [c++] stdlist::iterator it(l.begin()), end(l.end()); for(; it != end; ++it) ++(it->i_); Values (`test_class` and `itest_class`) and lists are created as explained in the first section. Note that for pointer lists, the iteration will suffer an extra indirection: they will need to access the `test_class` values stored in another container: [c++] stdptrlist::iterator it(l.begin()), end(l.end()); for(; it != end; ++it) ++((*it)->i_); These are the results: [table Write access times for Visual C++ 7.1 / Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [2031 / 8125] [1 / 1]] [[`safe_link` intrusive list] [2031 / 8281] [1 / 1.01]] [[`auto_unlink` intrusive list] [2031 / 8281] [1 / 1.01]] [[Standard list] [4218 / 10000] [2.07 / 1.23]] [[Standard compact pointer list] [4062 / 8437] [2.0 / 1.03]] [[Standard disperse pointer list] [8593 / 13125] [4.23 / 1.61]] ] [table Write access times for GCC 4.1.1 / MinGW over Windows XP [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [2343 / 8281] [1 / 1]] [[`safe_link` intrusive list] [2500 / 8281] [1.06 / 1]] [[`auto_unlink` intrusive list] [2500 / 8281] [1.06 / 1]] [[Standard list] [4218 / 10781] [1.8 / 1.3]] [[Standard compact pointer list] [3906 / 8281] [1.66 / 1]] [[Standard disperse pointer list] [8281 / 13750] [3.53 / 1.66]] ] [table Write access times for GCC 4.1.2 / Linux Kernel 2.6.18 (OpenSuse 10.2) [[Container] [Time in us/iteration (small object / big object)] [Normalized time (small object / big object)]] [[`normal_link` intrusive list] [2286 / 8468] [1 / 1.1]] [[`safe_link` intrusive list] [2381 / 8412] [1.04 / 1.09]] [[`auto_unlink` intrusive list] [2301 / 8437] [1.01 / 1.1]] [[Standard list] [3044 / 9061] [1.33 / 1.18]] [[Standard compact pointer list] [2755 / 7660] [1.20 / 1]] [[Standard disperse pointer list] [6118 / 12453] [2.67 / 1.62]] ] As with the read access test, the results show that intrusive containers outperform all other containers if the values are tightly packed in a vector. The disperse list is again the slowest. [endsect] [section:performance_results_conclusions Conclusions] Intrusive containers can offer performance benefits that cannot be achieved with equivalent non-intrusive containers. Memory locality improvements are noticeable when the objects to be inserted are small. Minimizing memory allocation/deallocation calls is also an important factor and intrusive containers make this simple if all objects to be inserted in intrusive containers are allocated using `std::vector` or `std::deque`. [endsect] [endsect] [section:release_notes Release Notes] [section:release_notes_boost_1_73_00 Boost 1.73 Release] * Fixed bugs: * [@https://github.com/boostorg/intrusive/issues/46 GitHub #46: ['UB due to union based type punning]] [endsect] [section:release_notes_boost_1_71_00 Boost 1.71 Release] * Fixed bugs: * [@https://github.com/boostorg/intrusive/pull/42 GitHub #42: ['Documentation does not describe treap priority_of_value changes]] * [@https://github.com/boostorg/intrusive/pull/43 GitHub #43: ['Fix tests with BOOST_INTRUSIVE_VARIADIC_TEMPLATES enabled]] * [@https://github.com/boostorg/intrusive/pull/45 GitHub #45: ['Disable variadic templates for MSVC-12 to avoid ICEs]] [endsect] [section:release_notes_boost_1_70_00 Boost 1.70 Release] * Fixed bugs: * [@https://github.com/boostorg/intrusive/pull/33 GitHub Pull #33: ['Fix compilation in case if key is void*, again]] * [@https://github.com/boostorg/intrusive/issues/34 GitHub Issue #34: ['-Wdeprecated-copy on gcc9]] * [@https://github.com/boostorg/intrusive/issues/35 GitHub Issue #35: ['key_of_value on treap_set seems to be broken in 1.69]] * [@https://github.com/boostorg/intrusive/issues/38 GitHub Issue #38: ['treap: Same type for priority and key comparison leads to ambiguous base class error]] * [@https://github.com/boostorg/intrusive/pull/39 GitHub Pull #39: ['Fix -Wextra-semi clang warnings]] [endsect] [section:release_notes_boost_1_67_00 Boost 1.67 Release] * Fixed bugs: * [@https://github.com/boostorg/intrusive/issues/29 GitHub Issues #29: ['Uninitialized variable warning pointer_plus_bits.hpp]] [endsect] [section:release_notes_boost_1_65_00 Boost 1.65 Release] * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/12894 Boost Trac #12894: ['Allow non std::size_t size_type]] * [@https://svn.boost.org/trac/boost/ticket/12698 Boost Trac #12698: ['base64 iterators can't be used with iterator_advance]] * [@https://github.com/boostorg/intrusive/pull/23 GitHub Pull #23: ['Conditionally replace deprecated/removed C++98 std::random_shuffle by...]] * [@https://github.com/boostorg/intrusive/pull/24 GitHub Pull #24: ['Adds support for MSVC ARM64 target]] [endsect] [section:release_notes_boost_1_64_00 Boost 1.64 Release] * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/12745 Boost Trac #12745: ['key_nodeptr_comp broken if the key type is void*]] * [@https://svn.boost.org/trac/boost/ticket/12761 Boost Trac #12761: ['intrusive::set::swap doesn't swap the comparison function*]] [endsect] [section:release_notes_boost_1_63_00 Boost 1.63 Release] * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/12556 Boost Trac #12556: ['member_value_traits.hpp has a missing #include]] [endsect] [section:release_notes_boost_1_62_00 Boost 1.62 Release] * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/11476 Boost Trac #11476: ['has_member_function_callable_with.hpp is massively broken with BOOST_NO_CXX11_DECLTYPE]] * [@https://svn.boost.org/trac/boost/ticket/11994 Boost Trac #11994: ['Support intrusive container key extractors that return the key by value]] * [@https://svn.boost.org/trac/boost/ticket/12184 Boost Trac #12184: ['clang -Wdocumentation warning]] * [@https://svn.boost.org/trac/boost/ticket/12190 Boost Trac #12190: ['Intrusive List + Flat Map combination crashes]] * [@https://svn.boost.org/trac/boost/ticket/12229 Boost Trac #12229: ['intrusive::unordered_set::rehash() broken]] * [@https://svn.boost.org/trac/boost/ticket/12245 Boost Trac #12245: ['bstree uses a shared static size_traits for constant_time_size]] * [@https://svn.boost.org/trac/boost/ticket/12432 Boost Trac #12432: ['Forced KeyOfValue creation when using custom compare on insert_check]] * Implemented `merge` functions in ordered associative containers. * Officially documented `root()` function for tree-based containers. [endsect] [section:release_notes_boost_1_61_00 Boost 1.61 Release] * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/11832 Boost Trac #11832: ['clang-cl + boost intrusive = miscompile]] * [@https://svn.boost.org/trac/boost/ticket/11865 Boost Trac #11865: ['Intrusive list explicit ctor error with Clang 3.6 (C++11/14)]] * [@https://svn.boost.org/trac/boost/ticket/11992 Boost Trac #11992: ['Add an overload of insert_check taking a key_type]] * [@https://github.com/boostorg/intrusive/pull/19 GitHub Pull #19: ['ebo_functor_holder: compile fix for copy constructor]] [endsect] [section:release_notes_boost_1_60_00 Boost 1.60 Release] * [link intrusive.advanced_lookups_insertions Advanced lookup and insertions] in ordered associative containers now support comparison functions that are not required to offer the same strict weak ordering as `key_compare`, the container must be partitioned in regards to the passed comparison object. * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/11701 Boost Trac #11701: ['Regression in boost::intrusive::set::equal_range]] * [@https://svn.boost.org/trac/boost/ticket/11765 Boost Trac #11765: ['sgtree.hpp:830: bad if test ?]] [endsect] [section:release_notes_boost_1_59_00 Boost 1.59 Release] * Implemented [link intrusive.map_multimap map and multimap-like interfaces]. * Refactored hashtable containers to reduce template instantiations. * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/11222 Boost Trac #11222: ['intrusive/pointer_traits.hpp fails to compile with C++98]] [endsect] [section:release_notes_boost_1_58_00 Boost 1.58 Release] * Reduced compile-time dependencies, headers, and the use of Boost.Preprocessor, specially for hooks and iterators. * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/6720 Boost Trac #6720: ['intrusive::unordered_set::clear_and_dispose does not compile on VC11 Beta when passed a stateless lambda]] * [@https://svn.boost.org/trac/boost/ticket/10771 Boost Trac #10771: ['remove_if is broken for slist]] * [@https://svn.boost.org/trac/boost/ticket/10853 Boost Trac #10853: ['problem with detection of const_cast_from]] * [@https://svn.boost.org/trac/boost/ticket/10987 Boost Trac #10987: ['bug in any_xxx_node_traits, returning by reference]] [endsect] [section:release_notes_boost_1_57_00 Boost 1.57 Release] * Experimental version of node checkers, contributed by Matei David. Many thanks! * Implemented [@http://www.open-std.org/JTC1/sc22/WG21/docs/papers/2013/n3644.pdf N3644: Null Forward Iterators] from C++14. * Fixed bugs: * [@https://github.com/boostorg/intrusive/pull/12 GitHub Pull #12: ['Fix MSVC14 warning C4456: declaration of 'x_parent_right' hides previous local declaration]] * [@https://svn.boost.org/trac/boost/ticket/10520 Boost Trac #10520: ['Conversion warning in intrusive/detail/utilities.hpp]] * [@https://svn.boost.org/trac/boost/ticket/10469 Boost Trac #10469: ['Erasing from intrusive unordered_multiset with optimize_multikey goes into an infinite loop]] [endsect] [section:release_notes_boost_1_56_00 Boost 1.56 Release] * Improved Doxygen generated reference and updated and fixed forward-declaration header. * [*ABI breaking]: Fixed ABI regression introduced in Boost 1.55 version, mainly noticeable on MSVC compilers. * [*Source breaking]: Removed previously deprecated `xxx_dont_splay` functions from splay containers, `splay_set_base_hook` and `splay_set_member_hook`from splay containers and `bool splay = true` extra parameter in `splaytree_algorithms` functions. * Fixed bugs: * [@https://svn.boost.org/trac/boost/ticket/8468 #8468: Compile error on visual studio 2010/2012 using vector with custom allocator and aligned types] * [@https://svn.boost.org/trac/boost/ticket/9332 #9332: ['"has_member_function_callable_with.hpp compile error on msvc-12.0"]]. * [@https://svn.boost.org/trac/boost/ticket/9650 #9650: ['"intrusive list with stateful value traits"]]. * [@https://svn.boost.org/trac/boost/ticket/9746 #9746: Modern Sun CC compiler detects error in intrusive library header] * [@https://svn.boost.org/trac/boost/ticket/9940 #9940: bad bug in intrusive list with safe_link (or auto_unlink) hooks] * [@https://svn.boost.org/trac/boost/ticket/9948 #9948: remove use of const_cast in intrusive containers] * [@https://svn.boost.org/trac/boost/ticket/9949 #9949: clear header node hooks upon intrusive container destruction] * [@https://svn.boost.org/trac/boost/ticket/9961 #9961: tests for hooks not derived frorm generic_hook] * Optimized tree rebalancing code to avoid redundant assignments. * Added 64 bit prime values for `suggested_upper_bucket_count`/`suggested_lower_bucket_count` in 64 bit platforms. * Deleted workarounds for old SUN_CC compilers, those are now unsupported as modern SunPro compilers are standard-corforming enough. [endsect] [section:release_notes_boost_1_55_00 Boost 1.55 Release] * [*Source breaking]: Deprecated `xxx_dont_splay` functions from splay containers. Deprecated `splay_set_base_hook` and `splay_set_member_hook`from splay containers, use `bs_set_base_hook` or `bs_set_member_hook` instead. Both will be removed in Boost 1.56. * [*ABI breaking]: Hash containers' end iterator was implemented pointing to one-past the end of the bucket array (see [@https://svn.boost.org/trac/boost/ticket/8698 #8698]) causing severe bugs when values to be inserted where allocated next to the bucket array. End iterator implementation was changed to point to the beginning of the bucket array. * Big refactoring in order to reduce template and debug symbol bloat. Test object files have been slashed to half in MSVC compilers in Debug mode. Toolchains without Identical COMDAT Folding (ICF) should notice size improvements. * Implemented [link intrusive.scary_iterators SCARY iterators]. [endsect] [section:release_notes_boost_1_54_00 Boost 1.54 Release] * Added `BOOST_NO_EXCEPTIONS` support (bug [@https://svn.boost.org/trac/boost/ticket/7849 #7849]). [endsect] [section:release_notes_boost_1_53_00 Boost 1.53 Release] * Fixed bugs [@https://svn.boost.org/trac/boost/ticket/7174 #7174], [@https://svn.boost.org/trac/boost/ticket/7529 #7529], [@https://svn.boost.org/trac/boost/ticket/7815 #7815]. * Fixed GCC -Wshadow warnings. * Added missing `explicit` keyword in several intrusive container constructors. * Replaced deprecated BOOST_NO_XXXX with newer BOOST_NO_CXX11_XXX macros. * Small documentation fixes. [endsect] [section:release_notes_boost_1_51_00 Boost 1.51 Release] * Fixed bugs [@https://svn.boost.org/trac/boost/ticket/6841 #6841], [@https://svn.boost.org/trac/boost/ticket/6907 #6907], [@https://svn.boost.org/trac/boost/ticket/6922 #6922], [@https://svn.boost.org/trac/boost/ticket/7033 #7033], * Added `bounded_range` function to trees. [endsect] [section:release_notes_boost_1_49_00 Boost 1.49 Release] * Fixed bugs [@https://svn.boost.org/trac/boost/ticket/6347 #6347], [@https://svn.boost.org/trac/boost/ticket/6223 #6223], [@https://svn.boost.org/trac/boost/ticket/6153 #6153]. [endsect] [section:release_notes_boost_1_48_00 Boost 1.48 Release] * Fixed bugs [@https://svn.boost.org/trac/boost/ticket/4797 #4797], [@https://svn.boost.org/trac/boost/ticket/5165 #5165], [@https://svn.boost.org/trac/boost/ticket/5183 #5183], [@https://svn.boost.org/trac/boost/ticket/5191 #5191]. [endsect] [section:release_notes_boost_1_46_00 Boost 1.46 Release] * Fixed bug [@https://svn.boost.org/trac/boost/ticket/4980 #4980], [endsect] [section:release_notes_boost_1_45_00 Boost 1.45 Release] * Added `function_hook` option. * Fixed bugs [@https://svn.boost.org/trac/boost/ticket/2611 #2611], [@https://svn.boost.org/trac/boost/ticket/3288 #3288], [@https://svn.boost.org/trac/boost/ticket/3304 #3304], [@https://svn.boost.org/trac/boost/ticket/3489 #3489], [@https://svn.boost.org/trac/boost/ticket/3668 #3668], [@https://svn.boost.org/trac/boost/ticket/3339 #3688], [@https://svn.boost.org/trac/boost/ticket/3698 #3698], [@https://svn.boost.org/trac/boost/ticket/3706 #3706], [@https://svn.boost.org/trac/boost/ticket/3721 #3721]. [@https://svn.boost.org/trac/boost/ticket/3729 #3729], [@https://svn.boost.org/trac/boost/ticket/3746 #3746], [@https://svn.boost.org/trac/boost/ticket/3781 #3781], [@https://svn.boost.org/trac/boost/ticket/3840 #3840], [@https://svn.boost.org/trac/boost/ticket/3849 #3849], [@https://svn.boost.org/trac/boost/ticket/3339 #3339], [@https://svn.boost.org/trac/boost/ticket/3419 #3419], [@https://svn.boost.org/trac/boost/ticket/3431 #3431], [@https://svn.boost.org/trac/boost/ticket/4021 #4021]. [endsect] [section:release_notes_boost_1_40_00 Boost 1.40 Release] * Code cleanup in bstree_algorithms.hpp and avl_tree_algorithms.hpp * Fixed bug [@https://svn.boost.org/trac/boost/ticket/3164 #3164]. [endsect] [section:release_notes_boost_1_39_00 Boost 1.39 Release] * Optimized `list::merge` and `slist::merge` * `list::sort` and `slist::sort` are now stable. * Fixed bugs [@https://svn.boost.org/trac/boost/ticket/2689 #2689], [@https://svn.boost.org/trac/boost/ticket/2755 #2755], [@https://svn.boost.org/trac/boost/ticket/2786 #2786], [@https://svn.boost.org/trac/boost/ticket/2807 #2807], [@https://svn.boost.org/trac/boost/ticket/2810 #2810], [@https://svn.boost.org/trac/boost/ticket/2862 #2862]. [endsect] [section:release_notes_boost_1_38_00 Boost 1.38 Release] * New treap-based containers: treap, treap_set, treap_multiset. * Corrected compilation bug for Windows-based 64 bit compilers. * Corrected exception-safety bugs in container constructors. * Updated documentation to show rvalue-references functions instead of emulation functions. [endsect] [section:release_notes_boost_1_37_00 Boost 1.37 Release] * Intrusive now takes advantage of compilers with variadic templates. * `clone_from` functions now copy predicates and hash functions of associative containers. * Added incremental hashing to unordered containers via `incremental<>` option. * Update some function parameters from `iterator` to `const_iterator` in containers to keep up with the draft of the next standard. * Added an option to specify include files for intrusive configurable assertion macros. [endsect] [section:release_notes_boost_1_36_00 Boost 1.36 Release] * Added `linear<>` and `cache_last<>` options to singly linked lists. * Added `optimize_multikey<>` option to unordered container hooks. * Optimized unordered containers when `store_hash` option is used in the hook. * Implementation changed to be exception agnostic so that it can be used in environments without exceptions. * Added `container_from_iterator` function to tree-based containers. [endsect] [endsect] [section:references References] * SGI's [@http://www.sgi.com/tech/stl/ STL Programmer's Guide]. [*Boost.Intrusive] is based on STL concepts and interfaces. * Dr. Dobb's, September 1, 2005: [@http://www.ddj.com/architect/184402007 ['Implementing Splay Trees in C++] ]. [*Boost.Intrusive] splay containers code is based on this article. * Olaf's original intrusive container library: [@http://freenet-homepage.de/turtle++/intrusive.html ['STL-like intrusive containers] ]. [endsect] [section:acknowledgements Acknowledgements] [*Olaf Krzikalla] would like to thank: * [*Markus Schaaf] for pointing out the possibility and the advantages of the derivation approach. * [*Udo Steinbach] for encouragements to present this work for boost, a lot of fixes and helpful discussions. * [*Jaap Suter] for the initial hint, which eventually lead to the member value_traits. [*Ion Gaztanaga] would like to thank: * [*Olaf Krzikalla] for the permission to continue his great work. * [*Joaquin M. Lopez Munoz] for his thorough review, help, and ideas. * [*Cory Nelson], [*Daniel James], [*Dave Harris], [*Guillaume Melquiond], [*Henri Bavestrello], [*Hervé Bronnimann], [*Kai Bruning], [*Kevin Sopp], [*Paul Rose], [*Pavel Vozelinek], [*Howard Hinnant], [*Olaf Krzikalla], [*Samuel Debionne], [*Stjepan Rajko], [*Thorsten Ottosen], [*Tobias Schwinger], [*Tom Brinkman] and [*Steven Watanabe] for their comments and reviews in the Boost.Intrusive formal review. * Thanks to [*Julienne Walker] and [*The EC Team] ([@http://eternallyconfuzzled.com]) for their great algorithms. * Thanks to [*Daniel K. O.] for his AVL tree rebalancing code. * Thanks to [*Ralf Mattethat] for his splay tree article and code. * Special thanks to [*Steven Watanabe] and [*Tobias Schwinger] for their invaluable suggestions and improvements. [endsect] [include auto_index_helpers.qbk] [section:index Indexes] [named_index class_name Class Index] [named_index typedef_name Typedef Index] [named_index function_name Function Index] [named_index macro_name Macro Index] [/index] [endsect] [xinclude autodoc.xml]