Summary:
This patch fixes http://llvm.org/PR31938. The description below is copy/pasted from the bug:
The standard says:
template<class charT, class traits = char_traits<charT>,
class Allocator = allocator<charT>>
class basic_string {
using value_type = typename traits::char_type;
// ...
basic_string(const charT* s, const Allocator& a = Allocator());
};
libc++ actually chooses to declare the constructor as
basic_string(const value_type* s, const Allocator& a = Allocator());
The implicit deduction guides from class template argument deduction make what was previously an implementation detail visible:
std::basic_string s = "foo"; // error, can't deduce charT.
The constructor in question is in the libc++ DSO, but fortunately it looks like fixing this will not result in an ABI break.
@rsmith How does this look? I did more than just the constructors mentioned in the PR, but IDK how far to take it.
Reviewers: mclow.lists, rsmith
Reviewed By: rsmith
Subscribers: cfe-commits, rsmith
Differential Revision: https://reviews.llvm.org/D29863
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@295393 91177308-0d34-0410-b5e6-96231b3b80d8
On Windows, we were using `Sleep` which is not alertable. This means
that if the thread was used for a user APC or WinProc handling and
thread::sleep was used, we could potentially dead lock. Use `SleepEx`
with an alertable sleep, resuming until the time has expired if we are
awoken early.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@295329 91177308-0d34-0410-b5e6-96231b3b80d8
clang cannot properly handle __declspec and __attribute__ on classes
right now. This prevents the shared_mutex tests from working. Disable
the use of the annotation on COFF targets.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294958 91177308-0d34-0410-b5e6-96231b3b80d8
The previous changes missed the change to include/cmath. These changes
allow some of the rand.distribution tests to pass on Windows.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294957 91177308-0d34-0410-b5e6-96231b3b80d8
Pull in the math functions from ucrt 14+ after auditing the library. It
seems that they are now complete for C99 math. Fixes more windows
tests!
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294918 91177308-0d34-0410-b5e6-96231b3b80d8
The newer versions of ucrt have the math routines. Use the CRT version
to determine if we should include the math routines. Fixes two tests
for Windows.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294899 91177308-0d34-0410-b5e6-96231b3b80d8
This recommits r294707 with additional fixes. The main difference is
libc++ now correctly builds without any ABI library.
exception.cpp is a bloody mess. It's full of confusing #ifdef branches for
each different ABI library we support, and it's getting unmaintainable.
This patch breaks down exception.cpp into multiple different header files,
roughly one per implementation. Additionally it moves the definitions of
exceptions in new.cpp into the correct implementation header.
This patch also removes an unmaintained libc++abi configuration.
This configuration may still be used by Apple internally but there
are no other possible users. If it turns out that Apple still uses
this configuration internally I will re-add it in a later commit.
See http://llvm.org/PR31904.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294730 91177308-0d34-0410-b5e6-96231b3b80d8
exception.cpp is a bloody mess. It's full of confusing #ifdef branches for
each different ABI library we support, and it's getting unmaintainable.
This patch breaks down exception.cpp into multiple different header files,
roughly one per implementation. Additionally it moves the definitions of
exceptions in new.cpp into the correct implementation header.
This patch also removes an unmaintained libc++abi configuration.
This configuration may still be used by Apple internally but there
are no other possible users. If it turns out that Apple still uses
this configuration internally I will re-add it in a later commit.
See http://llvm.org/PR31904.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294707 91177308-0d34-0410-b5e6-96231b3b80d8
The build was broken as there was no overload for long and
std::chrono::nanoseconds. Add an explicit conversion to use the
operator+.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294698 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Libcxx will define its own max_align_t when it is not available. However, the
availability checks today only check for Clang's definition and GCC's
definition. In particular, it does not check for musl's definition, which is the
same as GCC's but guarded with a different macro.
Reviewers: mclow.lists, EricWF
Reviewed By: EricWF
Subscribers: chandlerc, cfe-commits
Differential Revision: https://reviews.llvm.org/D28478
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294683 91177308-0d34-0410-b5e6-96231b3b80d8
A static assertion was misfiring since it checked
is_callable<Visitor, decltype(__variant_alt<T>.value)>. However
the decltype expression doesn't capture the value category as
required. This patch applies extra braces to decltype to fix
that.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294612 91177308-0d34-0410-b5e6-96231b3b80d8
Replace preprocess conditions of defined(_MSC_VER) &&
!defined(__clang__) with defined(_LIBCPP_MSVC). NFC.
Patch by Dave Lee!
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294171 91177308-0d34-0410-b5e6-96231b3b80d8
In addition to the PR for LWG 2773 this patch also ensures
that each of std::ignores constructors or assignment operators
are constexpr.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294165 91177308-0d34-0410-b5e6-96231b3b80d8
It is my opinion that libc++ should never use `<cassert>`, including in the `dylib`.
This patch remove all uses of `assert` from within libc++ and replaces most of them with `_LIBCPP_ASSERT` instead.
Additionally this patch turn `LIBCXX_ENABLE_ASSERTIONS` off by default,
because the standard library should not be aborting user programs unless explicitly asked to.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294107 91177308-0d34-0410-b5e6-96231b3b80d8
Visible definitions for basic_string::assign are sometimes emitted in
the dylib depending on the version of LLVM used to compile libc++.
This can cause the check-cxx-abilist target to fail.
This patch attempts marks the basic_string::assign templates as inline
to prevent this. That way the export list is consistent across LLVM
versions.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@294100 91177308-0d34-0410-b5e6-96231b3b80d8
While this change didn't really hurt, it does lead to spurious
warnings about not being able to override weak symbols if you end up
linking objects built with this change to ones built without it.
Furthermore, since __call_once_proxy is called indirectly anyway it
doesn't actually inline ever.
Longer term, it would probably make sense to give this symbol internal
visibility instead.
This reverts r291497
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@293581 91177308-0d34-0410-b5e6-96231b3b80d8
These member functions were decorated with `_LIBCPP_FUNC_VIS` when the
class is also decorated with external visibility. This breaks down when
building for PE/COFF, where the member function cannot be decorated if
it is within a decorated class. The class attribute will propagate to
the member. Remove the extraneous decoration.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@293454 91177308-0d34-0410-b5e6-96231b3b80d8
Microsoft's SAL has a `__deref` macro which results in a compilation
failure when building the filesystem module on Windows. Rename the
member function internally to avoid the conflict.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@293449 91177308-0d34-0410-b5e6-96231b3b80d8
This causes unnecessary warnings when building with `cl`. Newer
versions of the C standard permit the redefinition of the macro to the
same value (which is the case here), unfortunately, `cl` does not yet
implement this. Add a check to prevent the redefinition.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@293439 91177308-0d34-0410-b5e6-96231b3b80d8
Many thread-related libc++ test cases fail on FreeBSD, due to the
following -Werror warnings:
In file included from test/std/thread/thread.threads/thread.thread.this/sleep_until.pass.cpp:17:
In file included from include/thread:97:
In file included from include/__mutex_base:17:
include/__threading_support:222:1: error: mutex '__m' is still held at the end of function [-Werror,-Wthread-safety-analysis]
}
^
include/__threading_support:221:10: note: mutex acquired here
return pthread_mutex_lock(__m);
^
include/__threading_support:231:10: error: releasing mutex '__m' that was not held [-Werror,-Wthread-safety-analysis]
return pthread_mutex_unlock(__m);
^
include/__threading_support:242:1: error: mutex '__m' is still held at the end of function [-Werror,-Wthread-safety-analysis]
}
^
include/__threading_support:241:10: note: mutex acquired here
return pthread_mutex_lock(__m);
^
include/__threading_support:251:10: error: releasing mutex '__m' that was not held [-Werror,-Wthread-safety-analysis]
return pthread_mutex_unlock(__m);
^
include/__threading_support:272:10: error: calling function 'pthread_cond_wait' requires holding mutex '__m' exclusively [-Werror,-Wthread-safety-analysis]
return pthread_cond_wait(__cv, __m);
^
include/__threading_support:278:10: error: calling function 'pthread_cond_timedwait' requires holding mutex '__m' exclusively [-Werror,-Wthread-safety-analysis]
return pthread_cond_timedwait(__cv, __m, __ts);
^
6 errors generated.
This is because on FreeBSD, the pthread functions have lock annotations.
Since the functions in __thread_support are internal to libc++ only, add
no_thread_safety_analysis attributes to suppress these warnings.
Reviewers: mclow.lists, EricWF, delesley, aaron.ballman
Reviewed By: aaron.ballman
Subscribers: ed, aaron.ballman, joerg, emaste, cfe-commits
Differential Revision: https://reviews.llvm.org/D28520
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@293197 91177308-0d34-0410-b5e6-96231b3b80d8
This reverts commit r292883. Unfortunately <string_view> uses
_LIBCPP_ASSERT in a way which is not compatible with the C++11 dylib
build. I'll investigate more tomorrow.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@292923 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
It is my opinion that libc++ should never use `<cassert>`, including in the `dylib`. This patch remove all uses of `assert` from within libc++ and replaces most of them with `_LIBCPP_ASSERT` instead.
Additionally this patch turn `LIBCXX_ENABLE_ASSERTIONS` off by default, because the standard library should not be aborting user programs unless explicitly asked to.
Reviewers: mclow.lists, compnerd, smeenai
Reviewed By: mclow.lists
Subscribers: mgorny, cfe-commits
Differential Revision: https://reviews.llvm.org/D29063
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@292883 91177308-0d34-0410-b5e6-96231b3b80d8
<string> uses `decltype` in a way incompatible with `__typeof__`.
This is problematic when compiling <string> with Clang 3.4 because
even though it provides `__decltype` libc++ still used `__typeof__`
because clang 3.4 doesn't provide __is_identifier which libc++
uses to detect __decltype.
This patch manually detects Clang 3.4 and properly configures
for it.
git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@292833 91177308-0d34-0410-b5e6-96231b3b80d8
