As these are extensions and add extra requirements to the container elements,
they shouldn't be part of hash_table_data_* so that they won't get instantiated
if an unordered container is explicitly instantiated.
Merged revisions 46594-46604 via svnmerge from
https://svn.boost.org/svn/boost/branches/unordered/trunk
[SVN r46607]
On some compilers, friend functions are being instantiated when the main class
is explicitly instantiated. This is slightly problematic because the equality
functions (which are an extension) put extra requirements on the types used. So
I'm going to try defining the functions outside of the class, in the hope that
they won't get instantiated. If someone wants non-member functions to be
instantiated, I think it's reasonable to expect them to explicitly instantiate
them, especially as compilers don't seem to be consistent about this.
[SVN r46579]
'operator=' even when an 'operator=' which takes its argument by value has been
defined. This causes assignments to be ambiguous. To work around this, I'm
removing the definitions on those compilers - breaking move assignment, but
allowing other assignments to work.
I'm using 0x0593 because that's the earliest compiler that I've seen do the
right thing, but it might also work on earlier 5.9.x versions.
Refs #1923.
[SVN r45573]
hopefully better cross-platform support.
Merged revisions 44778-44835,44837-44918 via svnmerge from
https://svn.boost.org/svn/boost/branches/unordered/trunk
........
r44778 | danieljames | 2008-04-26 17:15:44 +0100 (Sat, 26 Apr 2008) | 2 lines
Remove a trailing comma.
........
r44779 | danieljames | 2008-04-26 17:23:51 +0100 (Sat, 26 Apr 2008) | 1 line
Merge in support for equality operators.
........
r44780 | danieljames | 2008-04-26 17:28:44 +0100 (Sat, 26 Apr 2008) | 1 line
Use my own list container to avoid working around STL container bugs.
........
r44833 | danieljames | 2008-04-28 08:03:43 +0100 (Mon, 28 Apr 2008) | 1 line
Better equality tests.
........
r44834 | danieljames | 2008-04-28 08:04:03 +0100 (Mon, 28 Apr 2008) | 1 line
Remove a superfluous check.
........
r44835 | danieljames | 2008-04-28 08:04:21 +0100 (Mon, 28 Apr 2008) | 1 line
Add equality reference documentation.
........
r44916 | danieljames | 2008-04-30 08:16:52 +0100 (Wed, 30 Apr 2008) | 1 line
New version of list.hpp
........
r44917 | danieljames | 2008-04-30 08:18:31 +0100 (Wed, 30 Apr 2008) | 1 line
Support compilers without ADL in the compile tests.
........
r44918 | danieljames | 2008-04-30 08:25:20 +0100 (Wed, 30 Apr 2008) | 7 lines
Change the typedef of buffered functions as it was confusing MSVC 6.5
get_allocator wasn't compiling when the allocator workaround is used because it
couldn't cast from the wrapped allocator to an allocator of another type. So
use value_alloc_ when it's available (it's only unavailable on compilers with
C++0x support, which don't require the workaround).
........
[SVN r44919]
make implementing emplace easier.
Merged revisions 44458-44460,44463-44465 via svnmerge from
https://svn.boost.org/svn/boost/branches/unordered/trunk
........
r44458 | danieljames | 2008-04-16 18:31:35 +0100 (Wed, 16 Apr 2008) | 2 lines
Pull out the buffered functions.
........
r44459 | danieljames | 2008-04-16 18:31:45 +0100 (Wed, 16 Apr 2008) | 4 lines
Inline construct_node and create_node into copy_group - these used to be used
in the implementation of insert but aren't now because of insert's exception
requirements, so keeping them around was just confusing.
........
r44460 | danieljames | 2008-04-16 18:31:54 +0100 (Wed, 16 Apr 2008) | 4 lines
Change link_node so that it takes a node_constructor containing a constructed
node instead of a node - this makes the code a little cleaner and also
simplifies exception safety.
........
r44463 | danieljames | 2008-04-16 18:35:11 +0100 (Wed, 16 Apr 2008) | 2 lines
Explicitly name the different insert overloads.
........
r44464 | danieljames | 2008-04-16 18:35:22 +0100 (Wed, 16 Apr 2008) | 2 lines
Explicitly name the different erase overloads.
........
r44465 | danieljames | 2008-04-16 18:35:33 +0100 (Wed, 16 Apr 2008) | 2 lines
Call the erase methods in hash_table_data directly.
........
[SVN r44489]
Merged revisions 44076-44414 via svnmerge from
https://svn.boost.org/svn/boost/branches/unordered/trunk
........
r44076 | danieljames | 2008-04-06 20:41:19 +0100 (Sun, 06 Apr 2008) | 1 line
Move semantics for compilers with rvalue references.
........
r44077 | danieljames | 2008-04-06 20:48:59 +0100 (Sun, 06 Apr 2008) | 1 line
Do move assignment 'properly'.
........
r44085 | danieljames | 2008-04-06 22:46:04 +0100 (Sun, 06 Apr 2008) | 1 line
Use normal references for the move members, reset the source buckets_ pointer to stop the buckets getting deleted, and remove a superflous pointer check.
........
r44109 | danieljames | 2008-04-07 23:49:36 +0100 (Mon, 07 Apr 2008) | 1 line
Add missing tests.
........
r44366 | danieljames | 2008-04-13 12:59:46 +0100 (Sun, 13 Apr 2008) | 1 line
Avoid using rvalue references in the implementation files.
........
r44368 | danieljames | 2008-04-13 15:13:33 +0100 (Sun, 13 Apr 2008) | 6 lines
Use a cut down version of the work in progress move library to implement move
semantics on more compilers. Unfortunately the move constructor with allocator
isn't really practical at the moment, since in the case where the container
can't be moved, and the allocators aren't equal it will copy the container
twice.
........
[SVN r44486]
Merged revisions 43922,43962,43966,43971,43981,43995-43996,44042,44046-44048,44057 via svnmerge from
https://svn.boost.org/svn/boost/branches/unordered/trunk
........
r43922 | danieljames | 2008-03-29 14:55:59 +0000 (Sat, 29 Mar 2008) | 1 line
Fix some typos in the reference documentation.
........
r43962 | danieljames | 2008-03-31 18:29:59 +0100 (Mon, 31 Mar 2008) | 1 line
Add a name variable to the release script, so that I can have different release names in different branches.
........
r43966 | danieljames | 2008-03-31 18:43:16 +0100 (Mon, 31 Mar 2008) | 1 line
Fix the image directory for standalone docs.
........
r43971 | danieljames | 2008-03-31 19:17:25 +0100 (Mon, 31 Mar 2008) | 1 line
Fix the unordered stylesheet.
........
r43981 | danieljames | 2008-04-01 13:31:26 +0100 (Tue, 01 Apr 2008) | 2 lines
Cast the pointer in the Visual C++ 6.5 _Charalloc method.
........
r43995 | danieljames | 2008-04-02 12:50:27 +0100 (Wed, 02 Apr 2008) | 1 line
Try using the interprocess containers for testing. Compilation is a bit slower but hopefully I'll run into less cross-platform problems.
........
r43996 | danieljames | 2008-04-02 13:25:49 +0100 (Wed, 02 Apr 2008) | 1 line
Revert my experiment with the interprocess containers. It didn't work out.
........
r44042 | danieljames | 2008-04-04 20:38:09 +0100 (Fri, 04 Apr 2008) | 1 line
Make hash table data a member of hash table, instead of a base.
........
r44046 | danieljames | 2008-04-05 12:38:05 +0100 (Sat, 05 Apr 2008) | 1 line
Remove rvalue_ref from Jamfile.v2 - I didn't mean to check it in.
........
r44047 | danieljames | 2008-04-05 12:39:38 +0100 (Sat, 05 Apr 2008) | 1 line
New constructors with allocators.
........
r44048 | danieljames | 2008-04-05 12:58:11 +0100 (Sat, 05 Apr 2008) | 1 line
Document the new constructors.
........
r44057 | danieljames | 2008-04-05 17:08:23 +0100 (Sat, 05 Apr 2008) | 1 line
Fix some bugs in the exception testing code.
........
[SVN r44417]
https://svn.boost.org/svn/boost/branches/unordered/trunk
........
r43840 | danieljames | 2008-03-24 17:25:07 +0000 (Mon, 24 Mar 2008) | 1 line
Fix a g++ warning.
........
r43844 | danieljames | 2008-03-24 17:56:28 +0000 (Mon, 24 Mar 2008) | 1 line
It's a new-ish year.
........
r43885 | danieljames | 2008-03-27 20:36:10 +0000 (Thu, 27 Mar 2008) | 1 line
The release script doesn't need to copy images and css - because that's now done in the jamfiles. Also tweak the shell script a tad bit.
........
r43890 | danieljames | 2008-03-27 23:01:40 +0000 (Thu, 27 Mar 2008) | 1 line
Starting to add a docbook bibliography.
........
r43894 | danieljames | 2008-03-27 23:24:18 +0000 (Thu, 27 Mar 2008) | 1 line
Redeclare 'data' in iterator_base to help compilers which have trouble with accessing the nested typedef.
........
[SVN r43895]
Copyright update.
Switch back to the version where the sentinel points to itself.
Remove alternative versions of swap.
Workaround a borland bug or two.
More consistent use of class/swap/template.
Avoid a few warnings.
Add a no-throw swap to the allocator for exception testing.
[SVN r3793]
structure for unordered_multimap/unordered_multiset that uses an extra
pointer per node but makes most operations more efficient when equivalent
nodes are present.
Now uses macros instead of template metaprogramming to implement the
differences between containers with equivalent and unique keys.
Removed the erase_iterator stuff which complicated matters more than it helped.
Now just using pointers with some helper functions.
[SVN r2950]
The standard doesn't add them for local iterators, but I'm guessing that's an
oversight. So for now, I've hidden them behind a preprocessor condition. Which
I'll probably remove before review.
[SVN r2944]
Several changes:
- Makes insert meet its exception specification (but needs some refactoring).
- Some Borland work arounds, but not all there yet, maybe never will be.
- Some extra doxygen comments, but not nearly enough.
- Possibly other things that I'm too lazy to look up.
[SVN r2642]
