Version 1.78.0
Version 1.78.0
December 8th, 2021 03:45 GMT
Platform | File | SHA256 Hash |
---|---|---|
unix | boost_1_78_0.tar.bz2 | 8681f175d4bdb26c52222665793eef08490d7758529330f98d3b29dd0735bccc |
boost_1_78_0.tar.gz | 94ced8b72956591c4775ae2207a9763d3600b30d9d7446562c552f0a14a63be7 | |
windows | boost_1_78_0.7z | 090cefea470bca990fa3f3ed793d865389426915b37a2a3258524a7258f0790c |
boost_1_78_0.zip | f22143b5528e081123c3c5ed437e92f648fe69748e95fa6e2bd41484e2986cc3 |
Known Issues
These are patches from library authors which were found too late to be fixed in the release.
New Libraries
-
No new libraries.
Updated Libraries
-
Asio:
-
Added an io_uring backend that may optionally be used for all I/O objects, including sockets, timers, and posix descriptors.
-
Added support for files, with new new classes for stream-oriented and random-access files. This feature currently supports I/O completion ports on Windows, and io_uring on Linux.
-
Added support for portable pipes on POSIX and Windows (when I/O completion ports are available).
-
Added support for registered buffers, which are optimised when using the io_uring backend.
-
Added experimental support for channels, that may be used to send and queue completions as messages.
-
Implemented improvements to
experimental::coro
. -
Disabled
aligned_alloc
on clang when using an MSVC runtime. -
Changed to use a faster implementation for
ip::network_v4::canonical()
. -
Added template specialisations for common uses to improve compile time.
-
Reduced the size of
io_context
executors to a single pointer. -
Increased the small object buffer size for
execution::any_executor
andany_io_executor
. -
Fixed multi-signature handling when variadic templates are disabled.
-
Fixed compatibility with new versions of gcc and clang.
-
Fixed compilation on Solaris.
-
Fixed defence against Qt-defined macros when building with MSVC.
-
Improved error messages emitted by
ssl
facilities. -
Fixed
bind_executor
compatibility with completion tokens. -
Fixed build errors when
BOOST_ASIO_USE_TS_EXECUTOR_AS_DEFAULT
is defined. -
Added missing move assignment operator to
awaitable<>
. -
Fixed an access violation when using coroutines with MSVC, due to incorrect alignment of allocated memory.
-
Fixed a cleanup issue in
experimental::parallel_group
that occured when the execution context was shut down with parallel operations still pending. -
Added clarifications to the documentation on concurrency hints.
-
Added documentation on error handling techniques for C++20 coroutines.
-
Fixed various warnings.
-
Consult the Revision History for further details.
-
-
Assert:
-
Added
source_location::to_string
.
-
-
Atomic:
-
Use process-local futex operations on Android for non-IPC waiting and notifying operations.
-
Added support for Linux targets that only define
SYS_futex_time64
syscall, such as riscv32. -
Added a workaround for incorrect result of
std::alignment_of
on clang 8 for 64-bit types on 32-bit x86 targets. -
Added a ulock backend for waiting and notifying operations on Darwin systems since Mac OS 10.12, iOS 10.0, tvOS 10.0 or watchOS 3.0.
-
On Windows, corrected discrepancy between atomic-type::always_has_native_wait_notify and the corresponding capability macros when targeting Windows 8 or later. The library will now directly use
WaitOnAddress
and related APIs from public headers and therefore require user to link withsynchronization.lib
if the user requires Windows 8 or later by definingBOOST_USE_WINAPI_VERSION
,_WIN32_WINNT
or similar macros. The library is linked automatically on compilers that support auto-linking (e.g. MSVC). -
Added support for types with padding bits, except unions, on compilers that provide a way to clear the padding bits. This feature is supported by gcc 11 and MSVC 14.2 (compiler version 19.27) and newer, as well as other compilers supporting similar intrinsics. On compilers that don't allow to clear the padding bits, types with padding are still generally not supported with the exception of 80-bit
long double
on x86 targets. A newBOOST_ATOMIC_NO_CLEAR_PADDING
capability macro is defined to indicate when clearing the padding is not supported. -
Initializing constructors of
atomic_ref
andipc_atomic_ref
no longer use atomic instructions to clear the padding bits in the referenced object. This reduces the cost of the atomic reference construction. -
Initializing constructors of
atomic
andipc_atomic
are nowconstexpr
for enums, classes and floating point types. For classes and floating point types, the constructors areconstexpr
if the compiler supportsconstexpr
std::bit_cast
, the type has no padding bytes and no padding is required to implement native atomic operations. -
In accordance with C++20, default constructors of
atomic
andipc_atomic
now perform value initialization of the contained object. For types without a user-defined default constructor, this means the default-constructed atomic will be zero-initialized. -
Added a workaround for compilation error on AIX caused by the assembler tool not supporting numeric labels. (PR#50)
-
Added a workaround for compilation error with Visual Studio 2015 prior to Update 3. (#52)
-
See changelog for more details.
-
-
Beast:
-
This maintenance update brings minor bug fixes and updated CI reporting.
-
Fixed security vulnerability CVE-2016-9840 in zlib implementation.
-
Fixed WebSocket permessage_deflate implementation. This should result in an increase in compression performance when using Beast WebSockets.
-
Removed the test stream and test framework's dependency on RTTI. You can now use our handy test stream to check your embedded asynchronous communications code!
-
We'd love to know how you or your company use Beast, consider adding an entry to the Companies and Individuals Using Beast list.
-
See the full Release Notes for a complete list of changes.
-
-
Core:
-
Added a generic implementation to
<boost/core/cmath.hpp>
, enabled whenBOOST_CORE_USE_GENERIC_CMATH
is defined or when the platform does not provide the necessary facilities in<cmath>
. -
Added
boost::core::type_name
, a utility function that returns the name of a type as a string. -
Added
boost::span
, a C++11 implementation of C++20'sstd::span
(Glen Fernandes).
-
-
Describe:
-
Added
has_describe_enumerators
,has_describe_bases
,has_describe_members
. -
Added
enum_to_string
,enum_from_string
. -
Added relational and stream insertion operators.
-
Added
descriptor_by_name
,descriptor_by_pointer
.
-
-
Filesystem:
-
v4:
path::filename
andpath::iterator
no longer return an implicit trailing dot (".") element if the path ends with a directory separator. Instead, an empty path is returned, similar to C++17 std::filesystem. This also affects other methods that are defined in terms of iterators or filename, such aspath::stem
,path::compare
orlexicographical_compare
. For example,path("a/b/") == path("a/b/.")
no longer holds true. (#193) -
v4:
path::lexically_normal
no longer produce a trailing dot (".") element and omits a directory separator after a trailing dot-dot ("..") element in the normalized paths. -
v4:
path
append operations now consider root name and root directory in the appended path. If the appended path is absolute, or root name is present and differs from the source path, the resulting path is equivalent to the appended path. If root directory is present, the result is the root directory and relative path rebased on top of the root name of the source path. Otherwise, the behavior is similar to v3. (#214) -
path::lexically_normal
now converts directory separators to preferred separators in the root name of the normalized paths. -
Optimized overloads of
path::assign
,path::append
,path::concat
and the corresponding operators to avoid unnecessary path copying and reduce the amount of code redundancy. -
On POSIX systems, fixed
absolute(p, base)
returning a path with root namebase.root_name()
ifp
starts with a root directory. In such a casep
is already an absolute path and should be returned as is. -
create_directories
no longer reports an error if the input path consists entirely of dot (".") and dot-dot ("..") elements. The implementation is no longer using recursion internally and therefore is better protected from stack overflow on extremely long paths. -
On Windows,
remove
now supports deleting read-only files. The operation will attempt to reset the read-only attribute prior to removal. Note that this introduces a possibility of the read-only attribute being left unset, if the operation fails and the original value of the attribute fails to be restored. This also affectsremove_all
. (#216) -
remove_all
now returnsstatic_cast< uintmax_t >(-1)
in case of error, similar to C++17 std::filesystem. -
Fixed a linking error about unresolved references to Boost.ContainerHash functions when user's code includes
boost/filesystem/path.hpp
but notboost/container_hash/hash.hpp
and the compiler is set to preserve unused inline functions. (#215) -
Added a workaround for MSVC and compatible compilers eliminating path globals cleanup in release builds. This could lead to a memory leak if Boost.Filesystem shared library was repeatedly loaded and unloaded in the process. (#217)
-
-
JSON:
-
Standalone mode of the library is deprecated.
-
Allow external libraries to forward declare
value_to
andvalue_from
. -
Fixed signed integer overflow in number parsing.
-
Add support for
/Zc:implicitNoexcept-
on MSVC.
-
-
Lambda2:
-
Added special cases in
operator<<
andoperator>>
when the first argument is a stream, to allowstd::cout << _1
. -
Added
operator->*
. -
Added
first
,second
.
-
-
Log:
-
General changes:
-
Updated detection of
std::codecvt
specializations forchar16_t
andchar32_t
for compatibility with C++17 and later. (PR#159) -
Added support for C++17
std::byte
type to receive methods of the inter-process message queue. -
On Windows, when building the library for Windows 8 or later, the library will use
nt62
tag in the version namespace to denote the target OS ABI. For example, the version namespace could be named asv2_mt_nt62
. This name will be part of all symbols exported by the library. Use theBOOST_USE_WINAPI_VERSION
macro consistenly when building Boost and your code to request the minimum target Windows version. -
Improved performance of SSSE3 and AVX2 implementations of the
dump
stream manipulator. -
Added support for CMake build system. Only library build is supported at this moment, Boost.Build is still used for running tests.
-
-
Bug fixes:
-
Fixed a race condition on POSIX systems in
reliable_message_queue::open_or_create
or the equivalent constructor. Multiple threads or processes callingopen_or_create
concurrently with sending or receiving messages from the same queue could end up corrupting the queue contents and potentially crashing the process. (#162) -
Added a workaround for
b2 install
failing on Windows because of missingboost_log_setup
library ifBOOST_LOG_WITHOUT_SETTINGS_PARSERS
is defined. (#164)
-
-
See changelog for more details.
-
-
Math:
-
Add support for cubic roots.
-
Add support for
constexpr
versions of various standard library math routines. -
Add support for Bezier polynomials.
-
Improve worst-case tanh-sinh integration performance.
-
Disable inadvertant use of integral types in integration routines.
-
Minor update for erf approximations when the result is with 2ulp of 1.
-
Allow Bernoulli code to be used on platforms with no atomic integers.
-
Improve the mode of the non-central Chi Squared distribution.
-
Improve certain edge cases for <cstdfloat> complex-valued elementary functions, see #507.
-
-
Multi-index Containers:
-
Added
contains
to key-based indices (issue #35). -
Added
merge
operations to key-based indices. The functionality goes beyond the standard specification for (unordered) associative containers in a number of ways, most notably:-
The source index can be of any type, including non key-based indices.
-
Partial merge is provided: for instance,
x.merge(y,first,last)
merges only the elements ofy
within [first
,last
).
-
-
Previous versions of
splice
for sequenced and random access indices were destructive, i.e. elements were copy-inserted into the destination and then erased from the source. Now,splice
is based on node transfer much asmerge
in key-based indices, and has been similarly extended to accept source indices of any type: in fact,splice
can be regarded as a frontend to the same functionality provided bymerge
in key-based indices. For reasons of backwards compatibility, the destructive behavior ofsplice
has been retained in the case that the source and destination containers have unequal allocators. -
The fact has been documented that index iterator types do only depend on
node_type
and the position of the index in itsmulti_index_container
, (except for hashed indices, where uniqueness/non-uniqueness is also a dependency). This has implications on the validity of iterators to elements transferred bymerge
orsplice
. This property is a variant of what has been called SCARY iterators in the C++ standard mailing lists. SCARYness is currently (August 2021) not mandated for standard containers. -
Iterator SCARYness is now also preserved in safe mode.
-
-
Multiprecision:
-
Rework rational_adaptor to remove dependency on Boost.Rational and improve performance.
-
Fix issue with type_traits and clang, see #383.
-
Fixed buggy Lehmer's GCD routine, see #370.
-
Fix mpc_complex constructors so that lossy conversions are explicit, see #363.
-
Correct miscellaneous configuration issues (Intel intrinsic usage and Apple clang).
-
Correct some iostream operators for unsigned char types.
-
Correct mpq_rational conversions to integer types, see #342.
-
-
Nowide:
-
Breaking Change:
do_unshift
for the UTF-32 specialization ofutf8_codecvt
now correctly returnsnoconv
instead ofok
-
Remove unnecessary branch in
utf8_codecvt
for UTF-16 output -
Suppress deprecation warnings in MSVC when using
utf8_codecvt
with unicode char types -
Fix possible double-free when setting a custom buffer (
setbuf
) afterfilebuf
already allocated an internal buffer
-
-
PFR:
-
Added missing
#include <memory>
for inclusion ofstd::addressof
-
Fixed -Wzero-length-array warning when dealing with empty aggregates
-
Fixed compilation on msvc compilers <= 1920 (thanks to Alexey Romanov aka @alexey-romanov)
-
Added basic CMakeLists.txt support (thanks to @pdimov aka Peter Dimov)
-
Multiple improvements for docs, including links to https://github.com/apolukhin/pfr_non_boost
-
Added misc/strip_boost_namespace.sh script to remove
boost::
andBOOST_
prefixes
-
-
Predef:
-
Version 1.13.1
-
Fix spelling of "epoch".
-
Add missing parenthesis in sparc.h (from tkoecker).
-
Update documentation to use Rouge code styling and Amber general style.
-
-
System:
-
Added support for source locations to
error_code
. -
Added
error_code::to_string
,error_condition::to_string
,error_code::what
. -
system_error::what()
now contains the source location, if present. -
Added
result<T, E = error_code>
, a class holding either a value or an error, defined in<boost/system/result.hpp>
.
-
-
Variant2:
-
Added
<boost/variant2.hpp>
. -
Added
unsafe_get<I>
. -
Added
visit_by_index
. -
Added
operator<<
.
-
Updated Tools
-
Build:
-
Includes release of B2 version 4.7.2.
-
Compilers Tested
Boost's primary test compilers are:
-
Linux:
-
Clang: 3.4, 3.5, 3.6, 3.7, 3.8, 3.9, 12.0.0
-
Clang, C++11: 3.4, 11.0.0
-
Clang, C++14: 3.5, 3.6, 3.7, 3.8, 3.9, 4.0, 5.0, 12.0.0
-
Clang, C++17: 6.0.1, 7.0.0, 8.0.0, 9.0.0, 10.0.0, 11.0.0, 12.0.0
-
Clang, C++20: 11.0.0, 12.0.0, 13.0.0
-
GCC: 4.6.3, 11
-
GCC, C++11: 4.7.3, 4.8.5, 11
-
GCC, C++14: 5.4.0, 6.4.0, 7.3.0, 8.0.1, 9.1.0, 11
-
GCC, C++17: 7.3.0, 8.0.1, 9.1.0, 11
-
GCC, C++20: 8.0.1, 9.1.0, 10, 11
-
-
OS X:
-
Apple Clang: 11.0.3
-
Apple Clang, C++11: 11.0.3
-
Apple Clang, C++14: 11.0.3
-
Apple Clang, C++17: 11.0.3
-
Apple Clang, C++20: 11.0.3
-
-
Windows:
-
Visual C++: 10.0, 11.0, 12.0, 14.0, 14.1, 14.2
-
Acknowledgements
Marshall Clow, Michael Caisse and Glen Fernandes managed this release.