Anotehr overhaul. Can now use `void_pointer` for links between nodes, although
it doesn't as I don't think `void_pointer` support is strong enough in existing
allocators.
Also no longer relies on using base pointers for custome pointer types. And
scaled back member function detection to just detect if an allocator has a
member, not what its signature is. I found that the trait could be confused by
ambiguous overloads. This might be fixable.
Better documentation of C++11 compliance to come.
[SVN r74859]
- Remove use of BOOST_DEDUCED_TYPENAME and BOOST_UNORDERED_PAIR_CAST, it's
unlikely that the compilers which require them will be able to cope with the
new version of unordered.
- Use the old equality algorithm if BOOST_UNORDERED_DEPRECATED_EQUALITY is
defined.
- Use SFINAE to control which overloads of `construct_impl` are available.
Fixes problems with differing overload resolution on different compilers.
- Support for piecewise pair construction.
- Only support the old variadic pair construction when
BOOST_UNORDERED_DEPRECATED_PAIR_CONSTRUCT is defined (also fixed some bugs).
- Avoid instantiating BOOST_RV_REF for non-classes.
- Support optional allocator member functions for compilers with SFINAE
expressions and Visual C++ 9.0/10.0
- Follow boost macro naming conventions.
- Improved portability for `allocator_traits` emulation.
Current compiler support:
- Full support for GCC 4.4+, Visual C++ 9.0+, Clang.
- All other compilers odn't support optional allocator members.
- No other errors for GCC 3.4.6+, Visual C++ 8.0, Intel, Pathscale.
- Visual Age has a compile error if `select_on_container_copy_construction`
isn't `const` (it should ignore it).
- `select_on_container_copy_construction` detection doesn't work on Sun.
- `unnecessary_copy_tests` is failling for vacpp on AIX, but not on linux.
- Warnings causing failures for Visual C++ with STLport and WM5.
[SVN r74234]
- Avoid using operator& with the value type.
- More comments in headers.
- Remove old clang workaround.
- Adjust use of inline to make Borland a little happier.
- Avoid `-Wconversion` warnings.
[SVN r67663]
https://svn.boost.org/svn/boost/trunk
........
r57126 | danieljames | 2009-10-24 12:56:30 +0100 (Sat, 24 Oct 2009) | 1 line
Update the intel compile flags.
........
r57139 | danieljames | 2009-10-24 18:53:03 +0100 (Sat, 24 Oct 2009) | 1 line
Fix unordered for intel strict.
........
r57150 | danieljames | 2009-10-25 10:54:28 +0000 (Sun, 25 Oct 2009) | 1 line
Fix the intel strict flag.
........
r57151 | danieljames | 2009-10-25 10:54:53 +0000 (Sun, 25 Oct 2009) | 1 line
Remove insert empty initializer lists, as there's a bug in gcc.
........
r57152 | danieljames | 2009-10-25 10:55:08 +0000 (Sun, 25 Oct 2009) | 1 line
Slightly rearrange the unordered container headers so that prev_prime is defined before it's used.
........
r57153 | danieljames | 2009-10-25 10:55:27 +0000 (Sun, 25 Oct 2009) | 1 line
Remove 'grouped' from hash_table as it isn't used and is a bit confusing.
........
[SVN r57179]
Merged revisions 57005-57006 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r57005 | danieljames | 2009-10-19 20:24:33 +0100 (Mon, 19 Oct 2009) | 6 lines
Use normal emplace implementation for emplace_hint and insert with hint.
There's a bug in the emplace_hint implementation for unordered
containers with equivalent keys. Since my tests missed it, I'm just
going to use the normal emplace implementation until I write better
tests.
........
r57006 | danieljames | 2009-10-19 20:32:09 +0100 (Mon, 19 Oct 2009) | 1 line
Fix allocator for construct from initializer list.
........
[SVN r57027]
Merged revisions 56441,56461,56468,56557-56562 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r56441 | danieljames | 2009-09-27 20:12:04 +0100 (Sun, 27 Sep 2009) | 1 line
Try supporting reference parameters in pairs. Probably not required.
........
r56461 | danieljames | 2009-09-29 00:06:03 +0100 (Tue, 29 Sep 2009) | 1 line
Remove the optimization for std::pair with a key reference. It'll be too much hassle to get a very unusual use case to work on all compilers.
........
r56468 | danieljames | 2009-09-29 08:46:44 +0100 (Tue, 29 Sep 2009) | 1 line
Just remove the test since the test itself doesn't work on most compilers.
........
r56557 | danieljames | 2009-10-03 17:40:26 +0100 (Sat, 03 Oct 2009) | 1 line
Fix the iterator category.
........
r56558 | danieljames | 2009-10-03 17:40:53 +0100 (Sat, 03 Oct 2009) | 2 lines
Update reference docs to latest version of draft standard and fill in
some missing details.
........
r56559 | danieljames | 2009-10-03 17:41:11 +0100 (Sat, 03 Oct 2009) | 1 line
Stricter insert exception tests.
........
r56560 | danieljames | 2009-10-03 17:41:32 +0100 (Sat, 03 Oct 2009) | 1 line
Insert using initializer lists.
........
r56561 | danieljames | 2009-10-03 17:42:00 +0100 (Sat, 03 Oct 2009) | 1 line
Update the unordered rationale.
........
r56562 | danieljames | 2009-10-03 17:42:20 +0100 (Sat, 03 Oct 2009) | 1 line
Make sure inserting from a range of types other than the value type is better tested.
........
[SVN r56700]
Merged revisions 55470,55877-55878,55901-55902,55921-55922,55990-55992,56009-56010,56329,56346-56349,56362-56363,56374 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r55470 | danieljames | 2009-08-08 19:50:00 +0100 (Sat, 08 Aug 2009) | 1 line
Remove empty svn:mergeinfo properties. This should reduce the amount of differences between trunk and release.
........
r55877 | danieljames | 2009-08-30 17:33:42 +0100 (Sun, 30 Aug 2009) | 1 line
Remove allocator_constructor since it's never used.
........
r55878 | danieljames | 2009-08-30 17:42:28 +0100 (Sun, 30 Aug 2009) | 6 lines
Initial checkin of new version of Boost.Unordered.
- More template use, less preprocessor use.
- Removed some of the Visual C++ 6 workarounds.
- Reduced memory use of the main object.
- Split into smaller headers.
........
r55901 | danieljames | 2009-08-31 11:39:25 +0100 (Mon, 31 Aug 2009) | 1 line
Detab.
........
r55902 | danieljames | 2009-08-31 11:39:40 +0100 (Mon, 31 Aug 2009) | 1 line
Remove unnecessary BOOST_DEDUCED_TYPENAMEs
........
r55921 | danieljames | 2009-08-31 16:33:28 +0100 (Mon, 31 Aug 2009) | 1 line
Remove a few unused parameters.
........
r55922 | danieljames | 2009-08-31 16:33:49 +0100 (Mon, 31 Aug 2009) | 1 line
Remove 'static' from next_node and node_count. Will hopefully make vacpp happy.
........
r55990 | danieljames | 2009-09-03 08:36:21 +0100 (Thu, 03 Sep 2009) | 1 line
Combine hash_structure and hash_table_manager.
........
r55991 | danieljames | 2009-09-03 08:37:14 +0100 (Thu, 03 Sep 2009) | 1 line
Remove some old Visual C++ workarounds.
........
r55992 | danieljames | 2009-09-03 08:37:30 +0100 (Thu, 03 Sep 2009) | 1 line
Add a small test to see if the tested compilers support out of line template methods.
........
r56009 | danieljames | 2009-09-04 08:02:28 +0100 (Fri, 04 Sep 2009) | 1 line
Fix link to n2691.
........
r56010 | danieljames | 2009-09-04 08:03:04 +0100 (Fri, 04 Sep 2009) | 1 line
Move size_ and cached_begin_bucket_ into table, rename hash_table_manager hash_buckets.
........
r56329 | danieljames | 2009-09-20 22:55:15 +0100 (Sun, 20 Sep 2009) | 2 lines
Since all the compilers support out of line template members use them
and lots of other things.
........
r56346 | danieljames | 2009-09-21 22:17:19 +0100 (Mon, 21 Sep 2009) | 1 line
Slightly more consistent variable names. In detail 'n' is now always a node pointer.
........
r56347 | danieljames | 2009-09-21 22:17:40 +0100 (Mon, 21 Sep 2009) | 1 line
Fix bug where container was reducing the number of buckets.
........
r56348 | danieljames | 2009-09-21 22:18:01 +0100 (Mon, 21 Sep 2009) | 1 line
Fix a bug that was causing unnecessary rehahes.
........
r56349 | danieljames | 2009-09-21 22:18:21 +0100 (Mon, 21 Sep 2009) | 1 line
Use std::max.
........
r56362 | danieljames | 2009-09-22 23:39:00 +0100 (Tue, 22 Sep 2009) | 1 line
Another std::max.
........
r56363 | danieljames | 2009-09-22 23:39:17 +0100 (Tue, 22 Sep 2009) | 1 line
Remove the emplace_hint implementation for unique containers as it isn't really used and seems to be causing sun 5.7 problems.
........
r56374 | danieljames | 2009-09-24 21:42:19 +0100 (Thu, 24 Sep 2009) | 1 line
Remove temporary test.
........
[SVN r56375]
Merged revisions 55099-55100,55132,55138,55184-55185 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r55099 | danieljames | 2009-07-22 23:37:52 +0100 (Wed, 22 Jul 2009) | 1 line
Fix the insert tests when there is a small number of buckets.
........
r55100 | danieljames | 2009-07-22 23:38:08 +0100 (Wed, 22 Jul 2009) | 1 line
Adjust the unordered defaults so that emplace takes more parameters and less buckets are created by default.
........
r55132 | danieljames | 2009-07-23 18:53:59 +0100 (Thu, 23 Jul 2009) | 1 line
Remove the emulation of single argument C++0x std::pair constructor.
........
r55138 | danieljames | 2009-07-23 23:17:20 +0100 (Thu, 23 Jul 2009) | 1 line
Try to work around an odd Visual C++ 8 bug.
........
r55184 | danieljames | 2009-07-26 19:59:33 +0100 (Sun, 26 Jul 2009) | 1 line
Some extra changelog notes.
........
r55185 | danieljames | 2009-07-26 20:00:40 +0100 (Sun, 26 Jul 2009) | 1 line
Update the reference documentation to mention that emplace is now emulated.
........
[SVN r55189]
Merged revisions 53505-53506,53525,53550,53552,53614 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r53505 | danieljames | 2009-05-31 16:50:56 +0100 (Sun, 31 May 2009) | 1 line
Disable incorrect Visual C++ 64-bit warnings. Ref #3082.
........
r53506 | danieljames | 2009-05-31 16:53:09 +0100 (Sun, 31 May 2009) | 1 line
Remove misplaced visual C++ warning pragma.
........
r53525 | danieljames | 2009-06-01 07:50:37 +0100 (Mon, 01 Jun 2009) | 1 line
Fix tests for when the library has support for initializer lists but the compiler doesn't.
........
r53550 | danieljames | 2009-06-01 20:17:49 +0100 (Mon, 01 Jun 2009) | 1 line
Get the type of the initializer_list right.
........
r53552 | danieljames | 2009-06-01 20:22:27 +0100 (Mon, 01 Jun 2009) | 1 line
Fix the unordered_map declaration in the tutorial. Fixes#3119.
........
r53614 | danieljames | 2009-06-03 23:48:49 +0100 (Wed, 03 Jun 2009) | 1 line
The move tests pass on 64 bit visual c++.
........
[SVN r53687]
Note that the previously rolled back commit was [53257], not [52357].
Merged revisions 52393-52394,52397,52884-52885,53255 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r52393 | danieljames | 2009-04-14 18:23:37 +0100 (Tue, 14 Apr 2009) | 2 lines
Implement full extract_key for compilers without SFINAE and variadic
templates.
........
r52394 | danieljames | 2009-04-14 18:23:51 +0100 (Tue, 14 Apr 2009) | 1 line
Use emplace instead of insert in the backend as it's more appropriate.
........
r52397 | danieljames | 2009-04-14 18:51:34 +0100 (Tue, 14 Apr 2009) | 1 line
Add stream output to the count test helper for unordered.
........
r52884 | danieljames | 2009-05-10 22:24:41 +0100 (Sun, 10 May 2009) | 19 lines
Cherrypick some unordered container changes from sandbox. Not including
anything which depends on the new move library.
------------------------------------------------------------------------
r52746 | danieljames | 2009-05-03 11:12:30 +0100 (Sun, 03 May 2009) | 1 line
Merge latest unordered container changes.
------------------------------------------------------------------------
r52747 | danieljames | 2009-05-03 11:15:35 +0100 (Sun, 03 May 2009) | 4 lines
Put the C++0x emplace implementations before the non-C++0x versions.
I'm going to change the non-C++0x to be macro heavy emulations of the
C++0x versions, so this will put the readable version first.
------------------------------------------------------------------------
r52748 | danieljames | 2009-05-03 11:15:44 +0100 (Sun, 03 May 2009) | 1 line
Refactor the unordered implementation a tad, to make implementing emplace less painful.
------------------------------------------------------------------------
........
r52885 | danieljames | 2009-05-10 22:25:09 +0100 (Sun, 10 May 2009) | 1 line
Merge emplace support for sandbox - but without move support.
........
r53255 | danieljames | 2009-05-25 20:45:06 +0100 (Mon, 25 May 2009) | 1 line
Unordered change log.
........
[SVN r53328]
* Support emplace for all compilers.
* Better configuration of C++0x features for when the appropriate headers
aren't available.
Merged revisions 52393-52394,52397,52884-52885,53127,53255 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r52393 | danieljames | 2009-04-14 18:23:37 +0100 (Tue, 14 Apr 2009) | 2 lines
Implement full extract_key for compilers without SFINAE and variadic
templates.
........
r52394 | danieljames | 2009-04-14 18:23:51 +0100 (Tue, 14 Apr 2009) | 1 line
Use emplace instead of insert in the backend as it's more appropriate.
........
r52397 | danieljames | 2009-04-14 18:51:34 +0100 (Tue, 14 Apr 2009) | 1 line
Add stream output to the count test helper for unordered.
........
r52884 | danieljames | 2009-05-10 22:24:41 +0100 (Sun, 10 May 2009) | 19 lines
Cherrypick some unordered container changes from sandbox. Not including
anything which depends on the new move library.
------------------------------------------------------------------------
r52746 | danieljames | 2009-05-03 11:12:30 +0100 (Sun, 03 May 2009) | 1 line
Merge latest unordered container changes.
------------------------------------------------------------------------
r52747 | danieljames | 2009-05-03 11:15:35 +0100 (Sun, 03 May 2009) | 4 lines
Put the C++0x emplace implementations before the non-C++0x versions.
I'm going to change the non-C++0x to be macro heavy emulations of the
C++0x versions, so this will put the readable version first.
------------------------------------------------------------------------
r52748 | danieljames | 2009-05-03 11:15:44 +0100 (Sun, 03 May 2009) | 1 line
Refactor the unordered implementation a tad, to make implementing emplace less painful.
------------------------------------------------------------------------
........
r52885 | danieljames | 2009-05-10 22:25:09 +0100 (Sun, 10 May 2009) | 1 line
Merge emplace support for sandbox - but without move support.
........
r53127 | danieljames | 2009-05-20 07:43:38 +0100 (Wed, 20 May 2009) | 1 line
Better configuration for boost.unordered.
........
r53255 | danieljames | 2009-05-25 20:45:06 +0100 (Mon, 25 May 2009) | 1 line
Unordered change log.
........
[SVN r53257]
Merged revisions 52658,52669,52673,52711 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r52658 | danieljames | 2009-04-29 11:05:17 +0100 (Wed, 29 Apr 2009) | 1 line
Fix the length of the prime number list.
........
r52669 | danieljames | 2009-04-29 22:43:41 +0100 (Wed, 29 Apr 2009) | 3 lines
Don't test prime_list::length on Visual C++.
Checking the array size doesn't seem to work on it.
........
r52673 | danieljames | 2009-04-30 06:08:40 +0100 (Thu, 30 Apr 2009) | 1 line
Revert changes to unordered, as the test fails on most compilers.
........
r52711 | danieljames | 2009-05-01 21:50:32 +0100 (Fri, 01 May 2009) | 5 lines
Use a preprocessor sequence for the primes.
This feels like overkill but it seems to be the most reliable way to ensure
that the length is correct. I obviously can't be trusted to get it right, and
the template hack seems to prevent Boost.Range from working.
........
[SVN r53043]
Merged revisions 52084,52245-52246,52304,52320,52323 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r52084 | danieljames | 2009-03-31 20:43:58 +0100 (Tue, 31 Mar 2009) | 1 line
Changelog for unordered and hash.
........
r52245 | danieljames | 2009-04-08 06:51:31 +0100 (Wed, 08 Apr 2009) | 5 lines
Detect gcc stdlib for gcc 4.0.1.
For some reason the normal macros aren't defined for the standard
library that comes with gcc 4.0.1 (although maybe just on BSDs?). So try
to detect the library for that compiler.
........
r52246 | danieljames | 2009-04-08 11:56:22 +0100 (Wed, 08 Apr 2009) | 7 lines
Include <utility> for some versions of gcc's library.
Sometimes gcc's <cstddef> doesn't define the C++ macros, so check for it
and include <utility> in that case. Also remove a workaround from
container_fwd.hpp
Fixes#2924.
........
r52304 | danieljames | 2009-04-10 20:25:32 +0100 (Fri, 10 Apr 2009) | 4 lines
Don't use debug containers on darwin.
I'm getting errors from the darwin 4.2 debug containers which appear to
a problem with its implementation.
........
r52320 | danieljames | 2009-04-11 08:53:59 +0100 (Sat, 11 Apr 2009) | 1 line
Don't need to include utility now that select_stdlib has been fixed.
........
r52323 | danieljames | 2009-04-11 09:26:20 +0100 (Sat, 11 Apr 2009) | 1 line
Also don't need to check for _GLIBCXX_CSTDDEF.
........
[SVN r52324]
Fixes#2756.
Merged revisions 51982-51983,51995 via svnmerge from
https://svn.boost.org/svn/boost/trunk
........
r51982 | danieljames | 2009-03-26 07:00:21 +0000 (Thu, 26 Mar 2009) | 3 lines
Revert [51409]
It isn't working on Borland.
........
r51983 | danieljames | 2009-03-26 07:00:46 +0000 (Thu, 26 Mar 2009) | 1 line
Try to destruct values in a way that all compilers might like.
........
r51995 | danieljames | 2009-03-26 21:09:51 +0000 (Thu, 26 Mar 2009) | 1 line
Give up and use another macro to destruct values.
........
[SVN r52065]
