Commit Graph

147 Commits

Author SHA1 Message Date
LeonineKing1199
4a42c93897 Fix erroneous usage of table::hash() impl which implicitly copy-constructs the const_key_type 2021-12-03 10:07:57 -08:00
LeonineKing1199
12977a50bc Add hopefully helpful comment to the source 2021-12-03 08:41:42 -08:00
LeonineKing1199
f252480bee Add missing formatting 2021-12-02 15:44:02 -08:00
LeonineKing1199
c9df887c4c Add member function template erase_key_unique_impl for usage in heterogeneous lookups 2021-12-02 15:38:07 -08:00
LeonineKing1199
03edf7f4a8 Add member function template find_previous_node_impl so it can be used in heterogenous contexts 2021-12-02 15:30:17 -08:00
Peter Dimov
5e8b6a9e55 Add comment describing the origin of mix64_policy::apply_hash 2021-11-29 06:14:20 +02:00
LeonineKing1199
8ba710637a Remove detail::make_dependent in favor of an extra template parameter in is_transparetn 2021-11-22 13:29:41 -08:00
LeonineKing1199
59db6cf788 Add SFINAE to transparent count() overload in unordered_map 2021-11-22 11:16:04 -08:00
Peter Dimov
c8abaf32ee Merge pull request #40 from LeonineKing1199/sfinae-updates
Update SFINAE expressions to be in the return type
2021-11-19 04:32:06 +02:00
LeonineKing1199
7ccd62ba98 Update SFINAE expressions to be in the return type instead of a defaulted function parameter 2021-11-18 12:54:17 -08:00
LeonineKing1199
2d69c7a5ca Add missing const-qualification of operator== for internal optional implementation 2021-11-18 10:19:20 -08:00
LeonineKing1199
a97160cf57 Fix bug in has_construct by using std::declval 2021-08-29 12:13:56 -07:00
Daniel James
e10fc0f93b Fix typo 2018-02-25 14:03:11 +00:00
Daniel James
09be9bae04 Use boost::is_nothrow_swappable, now that it's available everywhere 2018-01-27 09:38:19 +00:00
Daniel James
c01a8c4629 Oops, it's boost::to_address, no boost::to_pointer 2018-01-26 20:25:09 +00:00
Daniel James
ddf302fcc2 Use boost::to_pointer instead of pointer<..>::get
To get C++20 style recursive calls to `operator->`.
2018-01-26 18:59:16 +00:00
Daniel James
66533ace80 Add element_type to nodes, so that pointer_traits will work
Might be better to change the template signature of iterators, but that would
be a disruptive change to make at the moment.
2018-01-26 17:43:06 +00:00
Daniel James
f3476de893 Some explicit casts to avoid warnings on old GCC 2018-01-10 11:35:42 +00:00
Daniel James
d55c9565ab No template arguments in BOOST_MOVABLE_BUT_NOT_COPYABLE
Was causing errors on Visual C++ 7.1.
2018-01-10 09:58:08 +00:00
Daniel James
00a4185cf1 Reformat 2018-01-08 10:58:09 +00:00
Daniel James
ea28a3f98e Fix conversion to bool warning 2018-01-08 10:57:28 +00:00
Daniel James
34e54b35e8 Manually handle assigning hash/equality functions 2018-01-06 12:53:37 +00:00
Daniel James
f12009fc61 operator= noexcept support 2018-01-05 17:54:44 +00:00
Daniel James
5854090dc7 Swap noexcept support
Not properly supported as we don't have is_nothrow_swappable yet.
2018-01-05 17:48:13 +00:00
Daniel James
7e28fdd45a Make the current function_pair public 2018-01-05 17:10:13 +00:00
Daniel James
7615fabc80 Rewrite node handles using a lightweight limited optional
Will try to use std::optional when available. Also using
allocator_traits::is_always_equal support.
2018-01-03 23:15:55 +00:00
Daniel James
be0acc575f Implement allocator_traits::is_always_equal 2018-01-03 20:55:40 +00:00
Daniel James
9d558b010d Reformat 2017-12-28 11:44:57 +00:00
Daniela Engert
64441d2b64 Inheriting std::iterator is deprecated in c++17.
Therefore get rid of all of that and replace inheritance by lifting std::iterator's members into the derived class.

Signed-off-by: Daniela Engert <dani@ngrt.de>
2017-12-28 12:01:41 +01:00
Daniel James
92ce66be64 Rename 'bucket' variable to avoid shadow warning 2017-12-22 16:54:05 +00:00
Daniel James
15befe998e Rename table::get_bucket to get_bucket_pointer
All the other '*_bucket' methods return a bucket index.
2017-12-22 16:44:43 +00:00
Daniel James
b50e0d610f Initial implementation of template deduction guides 2017-12-19 12:56:51 +00:00
Daniel James
f99dee1917 Use predef for detecting piecewise construction
And detect it for recent dinkumware.
2017-12-19 12:11:36 +00:00
Daniel James
311e126ac4 Remove dependency on iterator 2017-12-03 18:58:52 +00:00
Daniel James
c037169e1a Update paths for headers moved from detail 2017-12-03 14:24:47 +00:00
Daniel James
978944fab2 Use same code for move constructing all containers
Copies the data layout from the source, so it doesn't need to hash anything.
2017-10-05 10:56:02 +01:00
Daniel James
705e69aefd Always call set_first_in_group
Probably don't need to, as we're using 0 for the first element in a group, but
it's quick so might as well.
2017-10-05 10:54:23 +01:00
Daniel James
e58081f6dc Drop some TODOs that are okay 2017-10-05 10:54:22 +01:00
Daniel James
4ac8a45a34 The max_load issue was fixed in the standard ages ago 2017-10-05 10:54:22 +01:00
Daniel James
6b5b968b97 Format with clang-format 4.0 2017-10-05 10:54:22 +01:00
Daniel James
f72b0353d4 Shuffle code around for readability
The new indentation made some of the code difficult to read, especially
where macros were concerned, so move things around and add more explicit
namespace declarations.
2017-06-11 20:55:59 +01:00
Daniel James
0676b4f4ca Change clang format indentation + .editorconfig file 2017-06-11 20:55:59 +01:00
Daniel James
5190a5d7f8 Stop dereferencing pointers to uninitialized memory
It's undefined behaviour. Still happens for piecewise construction
emultation for std::pair, I don't think there's anyway to avoid it.
I had considered using offsetof to get a pointer to a member, but that's
also undefined behaviour when a pair member doesn't have standard
layout. Piecewise construction emulation has other problems anyway.

So, this mostly fixes PR #5.

I also stopped using addressof in self-asssignment checks as operator&
is fine.
2017-06-04 08:47:02 +01:00
Daniel James
7775aa83df Decrease the limit for SunOS workaround
We have tests for 12.5 (5,20,0), and 12.5_next (5,21,0), I think both
are good enough to not require workarounds.
2017-05-25 08:53:34 +01:00
Daniel James
fc1604f2c8 Don't use allocator_traits::construct on GCC 4.6
Piecewise construction doesn't work uncopyable types.
2017-05-19 17:24:44 +01:00
Daniel James
4f5a2dabe9 Try to fix std::tuple on old Sun compilers
In order to use the workaround for both `boost::tuple` and `std::tuple`
the function would need to detect which was being used, in order to
decide whether to use `boost::tuples::length` or `std::tuple_size`.
Probably not difficult, but I don't have any way to test an
implementation.

So instead Just assume that if `std::tuple` is available it will work
without any workaround. Presumably once the compiler was able to support
`std::tuple` it will also support the necessary overloads.

I've left the version check as 5.21 so that failures will still show up
in the tests, but I'm sure it can be 5.20 and probably earlier.  Will
change before release.
2017-05-19 09:03:08 +01:00
Daniel James
ee9a5a2c77 Add a configuration macro for Sun C++ workarounds 2017-05-16 19:01:50 +01:00
Daniel James
6fffc738f7 Require std::tuple for full C++11 construction 2017-05-16 18:15:56 +01:00
Daniel James
12ee29579d Don't use full construction on Sun C++ w. libstdc++
There's an exception safety issue. Which is a pity as other than that it
seems fine. I'd assumed that support would be pretty good on all C++11
compilers, so I made it an 'all or nothing' feature, partly because
there are issues with pre-C++11 allocators, but this suggests partial
support might be desirable. Not sure I'll be able to (or want to) put
the time in though.
2017-05-12 21:58:52 +01:00
Daniel James
b6c6bfbe7f Statically handle 'is_unique' in assignments 2017-05-06 04:58:57 +01:00