diff options
Diffstat (limited to 'libs/intrusive/doc/intrusive.qbk')
| -rw-r--r-- | libs/intrusive/doc/intrusive.qbk | 21 |
1 files changed, 16 insertions, 5 deletions
diff --git a/libs/intrusive/doc/intrusive.qbk b/libs/intrusive/doc/intrusive.qbk index dcf97d318..4fac20735 100644 --- a/libs/intrusive/doc/intrusive.qbk +++ b/libs/intrusive/doc/intrusive.qbk @@ -1527,7 +1527,7 @@ 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, +[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 @@ -1577,7 +1577,7 @@ And they also can receive an additional option: [section:splay_set_multiset_example Example] -Now let's see a small example using +Now let's see a small example using [classref boost::intrusive::splay_set splay_set]/ [classref boost::intrusive::splay_multiset splay_multiset] containers: @@ -3261,8 +3261,8 @@ Thus [*Boost.Intrusive] offers its own `SCARY iterator` implementation, where it 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<>`, +* [*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<>`. @@ -3752,7 +3752,7 @@ The disperse list is again the slowest. 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 +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] @@ -3761,6 +3761,17 @@ to be inserted in intrusive containers are allocated using `std::vector` or `std [section:release_notes Release Notes] +[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! |