[boost] [interprocess] New snapshot in Boost Vault/CVS

From: Ion Gaztañaga (igaztanaga_at_[hidden])
Date: 2006-10-15 10:23:39

Hi to all,

  I've just uploaded a new Interprocess version. It's an ABI/Source 
breaking release that is a bit experimental, because the library has 
suffered many implementation changes:

*  Implemented N1780 proposal to LWG issue 233: Insertion hints in 
associative containers in interprocess boost::interprocess::multiset 
and boost::interprocess::multimap class.

* Source breaking: A shared memory object is now used including 
shared_memory_object.hpp header instead of shared memory.hpp.

* ABI breaking: Changed global mutex when initializing managed shared 
memory and memory mapped files. This change tries to minimize deadlocks.

[Note: This feature is experimental, and tries to avoid global 
Inteprocess deadlock when a process crashes holding the global lock. 
Now, the shared memory/memory mapped file contents are used to 
synchronize processes. Not sure if this change will be a step forward 
but, I need to know if users find it better than the old, global lock 
approach]

* Source breaking: Changed shared memory, memory mapped files and mapped 
region's open mode to a single boost::interprocess::mode_t type.

* Added extra WIN32_LEAN_AND_MEAN before including DateTime headers to 
avoid socket redefinition errors when using Interprocess and Asio in 
windows.

* ABI breaking: mapped_region constructor no longer requires classes 
derived from memory_mappable, but classes that match the MemoryMappable 
concept (interface).

* Added in-place reallocation capabilities to basic_string.

* ABI breaking: Reimplemented and optimized small string optimization. 
The narrow string class has zero byte overhead with an internal 11 byte 
buffer in 32 systems!

* Added move semantics to containers. Experimental and not documented 
yet. Improves performance when using containers of containers.

* ABI breaking: End nodes of node containers (list, slist, map/set) are 
now embedded in the containers instead of allocated using the allocator. 
This allows no-throw move-constructors and improves performance.

* ABI breaking: slist and list containers now have constant-time size() 
function. The size of the container is added as a member.

Future work:

-> I'm now working in Intrusive containers, adding non-raw pointer 
support. I want to base my container implementations in Intrusive, so I 
plan to pass Interprocess containers code to Intrusive code. My goal is 
to make sure that with Intrusive, everyone can construct containers easily.

-> I plan to optimize tree containers embedding red/black bit in the 
parent pointer, to reduce the overhead of a node from 4 words to 3 
words. I want to make this possible also with smart pointers.

-> I want to use Intrusive containers to program more efficient memory 
allocation algorithms and name-object indexes.

-> Implementation of process-shared sharable_mutex and convertible_mutex 
classes from N2094.

-> Bug fixes.

Regards,

Ion