Fixes the issues found by @zoecarver.
Only the unit tests don't fail for me. Will try to catch @zoecarver on IRC.

Fixes the issue found by @zoecarver .

Ping

Ping

Ping

Ping

Implemented the changes requested by @rsmith.

Add the proper markers in the unit test to update cxx_dr_status.html. As discussed on IRC; the up to date cwg_index.html is not public, so I only updated the unit test and removed the changes to cxx_dr_status.html.

Updated the unit tests as requested. This required the Sema::ActOnParamDefaultArgument to delay a part of the ODR validation until the default argument has been 'instantiated'.
As discussed on IRC; the up to date cwg_index.html is not public, so I only updated the unit test and removed the changes to cxx_dr_status.html.

Moved the testing from Parse to Sema.
Added additional safeguards for template instantiation.
Added more unit tests.
The comments may be a bit noisy, but they explain why the templates need to be tested at two locations.

Unfortunately I'm not able to quickly find the Doxygen output of Swift online. When I process:
A limited variant of \c @objc that's used for classes with generic ancestry.
with Doxygen I get:
A limited variant of that's used for classes with generic ancestry.
Since the @ is used for commands I think they should be escaped like:
include/swift/AST/Decl.h: /// such as \c \@testable and \c \@usableFromInline into account.

I think it should warn; according to the documentation [1] \c expects a word. Testing with Doxygen indeed gives a warning.

Thanks for the review.

Addresses @gribozavr comments.

All inline commands defined in include/clang/AST/CommentCommands.td require an argument. The escape commands, like \&, are handled in the switch starting at lib/AST/CommentLexer.cpp:366. They are stored as Text and not as an InlineCommand.

Remove the includes from the test.
Changed the std::is_constructible to is_same since the latter is easier to mock.

Fixes @lebedev.ri's remark and adds the requested test.
While working on the test I discovered the initial assumption in bug 35682 was incorrect. Updated the title and summary accordingly.

Thanks for the review. I don't have SVN access, can you commit these changes?

Addresses @rjmccall's remarks.
Fixes the tests for the nested lambda's.
As suspected the blocks also have the same nesting limit, thus added a test for them.

Thanks for feedback. I'll whether I can find a way to generate a nice test case.

Moved the test out of the loop as suggested by rjmccall.

tab -> space
adds an extra unit test for lambdas
fixes an off by one error found while testing the lambdas

It seems there already is a unit test "test/std/containers/associative/map/map.modifiers/insert_or_assign.pass.cpp". I'll add extra tests to this file.

Added more unit tests.
Moved the range validation from D62451 in this patch.

Removes the throw and move that to D62453.

I added 3 extra iterator positions for all hint cases:

• end() as you suggested.
• begin() this is most of the time the wrong iterator.
• the third element this is most of the time the wrong iterator. Just to avoid the code will use special optimizations on begin() and end().

Thanks for the review.

Thanks for the review!

Removed the ABI question in the comment

I added a unit test in part 3 of the series [1]. Would you prefer to have the test for this fix in this patch?

Thanks for the review.
Yes, please commit it for me since I don't have commit access.