Commit Graph

72 Commits

Author SHA1 Message Date
Daniel James
a1b1df84a0 Store bucket + whether first in group in node
Instead of the hash value.
2017-04-27 18:22:44 +01:00
Daniel James
408ebd0a0a Add node_bucket function 2017-04-27 18:22:44 +01:00
Daniel James
e9c4696544 Get rid of node_algo 2017-04-27 18:22:43 +01:00
Daniel James
ea64f2e46e Remove the grouped node stuff 2017-04-27 18:22:43 +01:00
Daniel James
ed326e2c87 Expand out more preprocessor repeats 2017-04-25 20:20:11 +01:00
Daniel James
bea4c6e29b undef some macros 2017-04-25 20:20:11 +01:00
Daniel James
61df9479e5 Rename unique_node 2017-04-25 20:20:11 +01:00
Daniel James
e3ab7b5d2e Remove BOOST_UNORDERED_EMPLACE_ARGSn macros
They're not used anywhere now.
2017-04-25 20:20:11 +01:00
Daniel James
679b73098e Option to use grouped nodes everywhere
Seems to work okay, but I'm not happy with and of the options. So I'm
going to look into doing something else before the next release.
2017-04-25 09:54:28 +01:00
Daniel James
cba643fc51 Make setting max_load_ a tad more consistent 2017-04-25 09:54:28 +01:00
Daniel James
1c8edf0298 Remove early exit for empty containers 2017-04-25 09:54:28 +01:00
Daniel James
b070bb5e49 Move everything from table_unique/table_equiv into table 2017-04-24 09:46:05 +01:00
Daniel James
e518120104 Rename some copy/assign/move methods
But with a hack so that the static casts will work until there's a
better method.
2017-04-24 09:46:05 +01:00
Daniel James
84f1ef6d2d Move constructor implementation into containers 2017-04-24 09:46:05 +01:00
Daniel James
77bf2b5e33 Rename methods to be different for unique/equiv keys
So that the implementation can be moved into a single class. Still some
other methods to rename. Some methods didn't need to be renamed (e.g.
try_emplace is only used with unique keys), but still renamed for
consistency.
2017-04-24 09:46:05 +01:00
Daniel James
fc08f62d6a Remove a TODO 2017-04-23 10:51:17 +01:00
Daniel James
7941771d61 Expand calls to at implementation 2017-04-23 10:09:18 +01:00
Daniel James
4f1c6e1ebf Expand calls to init/move_init 2017-04-23 10:09:18 +01:00
Daniel James
13ff1e7fb1 Expand calls to count and equal_range implementation 2017-04-23 10:09:18 +01:00
Daniel James
25b0b66e52 Expand calls to erase implementation
Doesn't work as well as the previous changes.
2017-04-23 10:09:18 +01:00
Daniel James
da835e88b8 Expand calls to insert_range 2017-04-23 10:09:18 +01:00
Daniel James
b6c229e2bb Expand calls to reserve implementation 2017-04-23 10:09:18 +01:00
Daniel James
19a45e028a Expand calls to generic_find_node 2017-04-23 10:09:18 +01:00
Daniel James
435b7450d4 Expand calls to max_size implementation 2017-04-23 10:09:18 +01:00
Daniel James
a41a0f3a06 Expand calls to load_factor implementation 2017-04-23 10:09:18 +01:00
Daniel James
814926ef31 Expand calls to clear implementation 2017-04-23 10:09:18 +01:00
Daniel James
f6f5ecdc00 Expand calls to emplace implementation
Also manually call the emplace macro up to 9 arguments, nicer error
messages for little effort.

Does it matter that there's no longer a nice backend for
`please_ignore_this_overload`? I don't think so, I was worried that it
would be confusing if triggered, but I'm not really aware of that ever
happening.
2017-04-23 10:09:18 +01:00
Daniel James
42b6b13943 Use the preprocessor to increment emplace limit
So that there's no need to add one.
2017-04-23 10:04:28 +01:00
Daniel James
cfe4c26f99 Explicitly write out boost::tuple template arguments
GCC 4.6 doesn't support using variadic arguments for a fixed length template.
There's a config macro for this, but might as well use the same code
everywhere.
2017-04-23 09:31:18 +01:00
Daniel James
3117611a55 New tuple implementation
Adjusts to use less arguments on Visual C++ 11, which will hopefully fix
it on that compiler. Also changed to be a little less preprocessor
heavy. I'm not sure about the __SUNPRO_CC support, hopefully recent
versions of that compiler will have better support, and can use the
normal implementation. Will check that later.
2017-04-22 18:31:10 +01:00
Daniel James
1e491533fa More consistent std::tuple configuration
Was getting a weird test failure for Visual C++ 11,
BOOST_NO_CXX11_HDR_TUPLE is defined, so the code doesn't support
std::tuple, but BOOST_UNORDERED_HAVE_PIECEWISE_CONSTRUCT was also
true, and so there are functions for constructing using
std::piecewise_construct/std::tuple, which don't work.

So, I'm assuming that if BOOST_UNORDERED_HAVE_PIECEWISE_CONSTRUCT is true,
then there must be a std::tuple. I guess it doesn't have full C++11 support,
which is why BOOST_NO_CXX11_HDR_TUPLE is defined, but it appears to be
good enough for us. If not, this will break things.
2017-04-21 20:32:56 +01:00
Daniel James
b067e65731 Clean table forward declarations 2017-04-20 22:59:00 +01:00
Daniel James
6bdf1ba244 Fix a comment 2017-04-19 10:21:13 +01:00
Daniel James
cee94e9fcb Fix unused parameter warning 2017-04-19 09:20:31 +01:00
Daniel James
e0054c7dd0 Remove alloc parameter from construct_value 2017-04-18 10:14:26 +01:00
Daniel James
3414e6628a Use allocator to construct/destroy nodes 2017-04-18 10:14:26 +01:00
Daniel James
08ce2c98e0 Rename call_construct to construct_value 2017-04-18 10:14:26 +01:00
Daniel James
6d79a322e2 Use macros to reduce call chain 2017-04-18 10:14:26 +01:00
Daniel James
9e70680044 Bypass construct_value/call_destroy_in a few places 2017-04-18 10:14:26 +01:00
Daniel James
7de8c91301 Remove calls to const_cast_pointer
It was needed because std::allocator_traits::construct doesn't work with
a const pointer (e.g. pointer to the first member of a std::pair). But
now we're only calling construct if BOOST_UNORDERED_CXX11_CONSTRUCTION
is true, so the allocator_traits::construct is no longer used here.
2017-04-18 10:14:26 +01:00
Daniel James
c333a7f9fc Use piecewise construction where possible 2017-04-18 10:14:26 +01:00
Daniel James
bc36a06a2d Comment about call_construct 2017-04-18 10:14:26 +01:00
Daniel James
e62ac22f0b Replace BOOST_UNORDERED_DETAIL_FULL_CONSTRUCT with BOOST_UNORDERED_CXX11_CONSTRUCTION
Require good construct support and piecewise construction. I don't know
if there are any platforms with good construct support, but no piecewise
construction, if there are then they'll no longer use
'allocator_traits::construct'/'allocator_traits::destruct'.
2017-04-18 10:14:26 +01:00
Daniel James
1092c972c9 Use allocator_traits to construct/destruct buckets 2017-04-18 10:14:26 +01:00
Daniel James
2b01bdbc25 More general deprecated check
`__has_cpp_attribute(deprecated)` returns true for C++11, but then warns
that it's a C++14 attribute, so only check in C++14 mode.
2017-04-16 16:34:22 +01:00
Daniel James
64a3be7d3e Use aligned_storage's address method to get pointer 2017-04-16 10:37:10 +01:00
Daniel James
ed8c0f9ecd Add deprecated attributes to deprecated functions 2017-04-15 17:35:09 +01:00
Daniel James
727e36e6a6 Comment about clear_buckets/create_buckets awkwardness.
They both leave the container in an invalid state that needs to be
cleaned up immediately. This confused me slightly.
2017-04-15 17:35:09 +01:00
Daniel James
09bddd8df4 Clear buckets before deleting nodes in clear.
The order shouldn't really matter as nothing throws an exception, but it
seems cleaner to never have any dangling pointers.
2017-04-15 17:35:08 +01:00
Daniel James
f089adc160 Change get_key to have node parameter 2017-04-15 17:35:08 +01:00