Boost.MultiIndex Release notes

Boost 1.34 release

Added random access indices.
Non key-based indices provide new rearrange facilities allowing for interaction with external mutating algorithms.
All predefined Boost.MultiIndex key extractors instantiated for a given type T can handle objects of types derived from or convertible to T (and chained pointers to those). Previously, only objects of the exact type specified (along with reference_wrappers and chained pointers to them) were accepted.
composite_key_compare and related classes accept operands not included in tuples as if they were passed in a tuple of length 1; this allows the user to omit tuple enclosing in lookup operations involving composite keys when only the first key is provided.
The core algorithms of ordered indices have been optimized, yielding an estimated reduction of about 5% in insertion times.
Size of ordered indices node headers have been reduced by 25% on most platforms, using a well known optimization technique.
The tutorial has been restructured, new examples added.
Maintenance fixes.

For ordered and hashed indices, erase(it) and erase(first,last) now return an iterator to the element following those being deleted (previously nothing was returned), in accordance with the C++ Standard Library Defect Report 130 and issue 6.19 of TR1 Issues List.
Boost.MultiIndex offers the usual guarantees with respect to multithreading code provided by most STL implementations:
1. Concurrent access to different containers is safe.
2. Concurrent read-only access to the same container is safe.
In previous versions of the library, the latter guarantee was not properly maintained if the safe mode was set. This problem has been fixed now.
Maintenance fixes.

Added hashed indices, whose interface is based on the specification for unordered associative containers by the C++ Standard Library Technical Report (TR1).
Added serialization support for Boost.Serialization.
Destruction of multi_index_containers and clear memfuns now perform faster.
Internal changes aimed at reducing the length of symbol names generated by the compiler; cuts of up to a 50% can be achieved with respect to the Boost 1.32 release. This results in much shorter and more readable error messages and has also a beneficial impact on compilers with strict limits on symbol name lengths. Additionally, a section on further reduction of symbol name lengths has been added.
Restructured some parts of the documentation, new examples.
Maintenance fixes.

Revised February 6th 2006