Commit Graph

1248 Commits

Author SHA1 Message Date
Daniel James
9119a42b7d Factor insert_exception_tests for better code reuse 2017-05-07 18:47:24 +01:00
Daniel James
8af4b37d14 Rewrite insert exception tests
Not going to do this for all the tests. It's more effort than I
expected.
2017-05-07 17:54:34 +01:00
Daniel James
31c5b5bfa1 Merge exception tests 2017-05-07 17:54:34 +01:00
Daniel James
b6c6bfbe7f Statically handle 'is_unique' in assignments 2017-05-06 04:58:57 +01:00
Daniel James
6e074d7165 Get rid of delete_nodes 2017-05-06 04:58:57 +01:00
Daniel James
0489069419 Use 'limited_range' to catch error in exception tests
This would have caught the error fixed in 3fe259a79e.
2017-05-06 04:47:59 +01:00
Daniel James
597eb5a3fd Repeat the assign/insert exception tests a few times 2017-05-06 04:47:59 +01:00
Daniel James
3fe259a79e Fix creating exception::less from exception::equal_to 2017-05-05 12:58:26 +01:00
Daniel James
6ef17a0f0e Remember to disable exceptions before checking final value 2017-05-05 00:46:07 +01:00
Daniel James
47a8c3fc67 Fix exception handling in rehash_impl
And improve tests so they will catch the error, and other similar errors.
2017-05-04 19:30:18 +01:00
Daniel James
d49d0e90a8 Delete nodes directly instead of through previous node 2017-05-04 00:14:08 +01:00
Daniel James
77bd45b1fa Make second parameter of delete_nodes a node_pointer 2017-05-04 00:14:05 +01:00
Daniel James
622dff50df Fix some code that could be using next_node 2017-05-03 23:36:09 +01:00
Daniel James
41f6a051ef Some more configuration comments 2017-05-03 04:21:53 +01:00
Daniel James
d05619095c Workaround problems with forward_as_tuple in older versions of clang 2017-05-03 04:21:52 +01:00
Daniel James
2f8492d720 Fix libc++ configuration 2017-05-03 04:21:52 +01:00
Daniel James
7911f491f6 Try using own allocator_traits with Sun compiler
The good news is that the old Sun workarounds aren't needed any more.
Unfortunately, there are a lot of exception test errors for
unordered_map and unordered_multimap when using libstdc++, which
probably means that std::pair isn't exception safe, which is a bit odd.
But first try using our allocator_traits implementation instead of the
standard one to see if that makes a difference. If it doesn't then I'll
probably just disable C++11 construction on this compiler, which should
fix the problem but will make allocators less useful.
2017-05-03 04:21:52 +01:00
Daniel James
d84a57441b Use the unordered container for the fake merge in merge test
The merge tests into containers with unique keys are failing on some
platforms. My guess is that because of differences between ordering of
nodes with equivalent keys that different nodes are being 'merged' in
the unordered containers and the tracker containers. So when creating
the fake merge, use the unordered container as a reference. This is a
little less pure, but should be a good enough test.
2017-05-01 22:16:13 +01:00
Daniel James
b8c754d230 Set the high bit in bucket_info_ to false for first node in group
And true for the reset, so that in containers with unique keys
bucket_info_ is always just the bucket index.
2017-05-01 21:03:11 +01:00
Daniel James
338a94e577 Better rvalue emulation support in extractors
Means that inserting rvalues into unordered_set/unordered_map won't
create a node if no insert is required.
2017-05-01 21:03:11 +01:00
Daniel James
2e14c340a8 Reformat 2017-05-01 21:03:11 +01:00
Daniel James
35522d3ee0 Fix merging between containers with different hash/equality
This reverts commit 20b0c0a6d8.
2017-04-30 14:29:05 +01:00
Daniel James
8c139940e7 More merge tests between compatible containers 2017-04-30 14:29:05 +01:00
Daniel James
7b5f73f6c2 Disable all sunpro workarounds on latest version
I have no idea if they're still working. I wasn't able to run the latest
solaris in a virtual machine on my computer, so this is the only way to
test.
2017-04-30 10:44:54 +01:00
Daniel James
e7a3487df4 Remove policy template parameter from local iterators 2017-04-30 10:44:54 +01:00
Daniel James
c243895fc0 Remove odd check that should never be true
I think it was left over from the old grouped node implementation.
2017-04-30 10:44:54 +01:00
Daniel James
451d0f2fc5 Constructing nodes is nothrow, so no need to track 2017-04-30 10:43:06 +01:00
Daniel James
c75b332240 Cleaner create_buckets implementation 2017-04-30 10:43:06 +01:00
Daniel James
899248acbf Avoid shadow warning on older versions off GCC 2017-04-29 09:31:17 +01:00
Daniel James
cae6b121b2 Improve test coverage a little 2017-04-28 21:26:21 +01:00
Daniel James
5f6ee3da9c Use coveralls 2017-04-28 09:53:50 +01:00
Daniel James
bfcdd51b4a Fix try_emplace overload 2017-04-28 09:53:50 +01:00
Daniel James
10b736d407 Remove BOOST_UNORDERED_CALL_CONSTRUCT0 2017-04-28 09:53:50 +01:00
Daniel James
0b61e6defb Stop using allocators to construct/destroy internal types
The standard doesn't allow it. I should have known that.
2017-04-27 19:02:10 +01:00
Daniel James
28f529100d Merge branch 'feature/simpler-data-structure' into develop
C++17 requires that unordered_map has the same type of node as
unordered_multimap, and that unordered_set has the same type of node as
unordered_multiset. This didn't seem particularly useful to me and
contradicts the old implementation which had different nodes, I put a
lot of effort into trying to abstract out the difference and make it
selectable using a macro, so that the old implementation would still by
available for anyone who doesn't care about strict compatibility.

But I think that was a mistake, it was making things too complicated and
for too little gain. The default would still be inefficient containers
for equivalent keys, and using the macro could lead to problems down the
line.

So I've switched to using a much simpler implementation which just marks
the first node in a group of equivalent nodes. This isn't as fast when
there are a lot of elements with equivalent keys - it can't skip to the
end of a group of nodes, but at least it avoids having to do a lot of
potentially expensive comparisons.

It's also a lot closer to the intent of the standard, even if I disagree
with that intent.
2017-04-27 18:22:53 +01:00
Daniel James
6466ce0b51 Make table the same for unique and equivalent keys 2017-04-27 18:22:44 +01:00
Daniel James
03baef8b28 Remove Types::is_unique 2017-04-27 18:22:44 +01:00
Daniel James
f1435d53d4 Remove 'init' method from nodes 2017-04-27 18:22:44 +01:00
Daniel James
20b0c0a6d8 Only consider one node from each group in merge_unique 2017-04-27 18:22:44 +01:00
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
94a3a9baf9 Fix some quickbook markup 2017-04-25 21:14:11 +01:00
Daniel James
da27ae4de6 Group together similar overloads of try_emplace 2017-04-25 20:20:11 +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