Release open for bug fixes and documentation updates. Other changes by permission of a release manager. Next milestone: Boost 1.93.0 closed Wednesday, December 2 Calendar: https://www.boost.org/calendar/ — Marshall :
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
Hi, I am requesting permission to merge a fix for a regression in the noncentral_t_distribution in boost.math: https://github.com/boostorg/math/pull/1639. This commit has cycled green against both Develop and Master. The issue was reported after the merge window closed: https://github.com/boostorg/math/issues/1637. Matt
El 12/11/2026 a las 14:02, Matt Borland via Boost escribió:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
Hi,
I am requesting permission to merge a fix for a regression in the noncentral_t_distribution in boost.math: https://github.com/boostorg/math/pull/1639. This commit has cycled green against both Develop and Master. The issue was reported after the merge window closed: https://github.com/boostorg/math/issues/1637.
Matt
Hi Matt, You don't need permimssion for small bug fixes and documentation changes. Best, Ion
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Integer develop to master: https://github.com/boostorg/integer/compare/master...develop The commits fix signed integer overflow and infinite recursion in gcd() for various corner cases of input arguments involving signed minimum values (e.g. INT_MIN).
El 12/11/2026 a las 21:45, Andrey Semashev via Boost escribió:
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Integer develop to master:
https://github.com/boostorg/integer/compare/master...develop
The commits fix signed integer overflow and infinite recursion in gcd() for various corner cases of input arguments involving signed minimum values (e.g. INT_MIN).
Go ahead. Bug fixes and small fixes (e.g. like warning removal) don't need permission. Best, Ion
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Thread develop to master: https://github.com/boostorg/thread/compare/master...develop This fixes TLS destructors and thread exit handlers not working on Windows ARM64X target. There are also test changes to improve CI reliability for timed tests.
El 13/11/2026 a las 12:29, Andrey Semashev via Boost escribió:
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Thread develop to master:
https://github.com/boostorg/thread/compare/master...develop
This fixes TLS destructors and thread exit handlers not working on Windows ARM64X target. There are also test changes to improve CI reliability for timed tests.
I see Appveyor cygwin/mingw tests have failed: https://ci.appveyor.com/project/viboes/thread/builds/55079416 Is this expected? Best, Ion
On 14 Nov 2026 01:34, Ion Gaztañaga wrote:
El 13/11/2026 a las 12:29, Andrey Semashev via Boost escribió:
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Thread develop to master:
https://github.com/boostorg/thread/compare/master...develop
This fixes TLS destructors and thread exit handlers not working on Windows ARM64X target. There are also test changes to improve CI reliability for timed tests.
I see Appveyor cygwin/mingw tests have failed:
https://ci.appveyor.com/project/viboes/thread/builds/55079416
Is this expected?
Sort of. Timing-related tests are unreliable in Boost.Thread, and I can't seem to win this battle no matter how I try. Cygwin seems to fail in one of those tests. One TLS test took too long to finish and left some of the spawned threads running, which then interfered with the following tests. Cygwin on GitHub Actions have passed. I've increased the TLS test timeouts. MinGW gcc-6.3.0 is broken because it is missing synchronization.lib, which is required by Boost.Atomic. I've replaced it with gcc 7.3.0. The CI with my changes is running now.
El 14/11/2026 a las 0:29, Andrey Semashev via Boost escribió:
On 14 Nov 2026 01:34, Ion Gaztañaga wrote:
El 13/11/2026 a las 12:29, Andrey Semashev via Boost escribió:
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Thread develop to master:
https://github.com/boostorg/thread/compare/master...develop
This fixes TLS destructors and thread exit handlers not working on Windows ARM64X target. There are also test changes to improve CI reliability for timed tests.
I see Appveyor cygwin/mingw tests have failed:
https://ci.appveyor.com/project/viboes/thread/builds/55079416
Is this expected?
Sort of. Timing-related tests are unreliable in Boost.Thread, and I can't seem to win this battle no matter how I try. Cygwin seems to fail in one of those tests. One TLS test took too long to finish and left some of the spawned threads running, which then interfered with the following tests. Cygwin on GitHub Actions have passed. I've increased the TLS test timeouts.
MinGW gcc-6.3.0 is broken because it is missing synchronization.lib, which is required by Boost.Atomic. I've replaced it with gcc 7.3.0.
The CI with my changes is running now.
Ok, thanks. If it looks ok, please go ahead and merge it. Ion
On Thu, Nov 12, 2026 at 3:36 AM Marshall Clow via Boost <boost@lists.boost.org> wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Interval develop to master: https://github.com/boostorg/interval/compare/master...develop The commit fixes construction of numeric intervals from integers in some cases. Previously, the constructed intervals could be unreasonably wide or not include the original value due to incorrect rounding.
El 14/11/2026 a las 20:24, Andrey Semashev via Boost escribió:
On Thu, Nov 12, 2026 at 3:36 AM Marshall Clow via Boost <boost@lists.boost.org> wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Interval develop to master:
https://github.com/boostorg/interval/compare/master...develop
The commit fixes construction of numeric intervals from integers in some cases. Previously, the constructed intervals could be unreasonably wide or not include the original value due to incorrect rounding.
Go ahead. Ion
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I am requesting to merge this PR in Decimal to Master: https://github.com/boostorg/decimal/pull/1560 . It contains a fix for a new -Werror=sign-conversion warning that showed up in Boost superproject CI with the recent switch in ubuntu-latest on Github Actions to Ubuntu 26.04. This fix cycles green against Develop and Master. Matt
El 15/11/2026 a las 14:43, Matt Borland via Boost escribió:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I am requesting to merge this PR in Decimal to Master: https://github.com/boostorg/decimal/pull/1560 . It contains a fix for a new -Werror=sign-conversion warning that showed up in Boost superproject CI with the recent switch in ubuntu-latest on Github Actions to Ubuntu 26.04. This fix cycles green against Develop and Master.
Matt
Seems pretty safe, all checks passed, so go ahead. Best Ion
I'd like to merge Boost LEAF develop into master. It's just bug fixes and documentation fixes. Of course I'll wait for GHA to succeed for develop, which it should since it has already succeeded on this git hash. On Wed, Nov 11, 2026 at 7:36 PM Marshall Clow via Boost < boost@lists.boost.org> wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
Next milestone: Boost 1.93.0 closed Wednesday, December 2
Calendar: https://www.boost.org/calendar/
— Marshall :
_______________________________________________ Boost mailing list -- boost@lists.boost.org To unsubscribe send an email to boost-leave@lists.boost.org https://lists.boost.org/mailman3/lists/boost.lists.boost.org/ Archived at: https://lists.boost.org/archives/list/boost@lists.boost.org/message/ZQNQGFGY...
El 16/11/2026 a las 3:09, Emil Dotchevski via Boost escribió:
I'd like to merge Boost LEAF develop into master. It's just bug fixes and documentation fixes. Of course I'll wait for GHA to succeed for develop, which it should since it has already succeeded on this git hash.
Fine, go ahead. Ion
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I am requesting to merge a small bugfix to master on Boost.Math (https://github.com/boostorg/math/pull/1659). iround has a forward declaration marked with BOOST_MATH_GPU_ENABLED, but the function definition did not. Previously, this annotation was implicitly added in CUDA, but that no longer is the case so an error occurs when iround is used in CUDA kernels. Best, Matt
El 18/11/2026 a las 14:37, Matt Borland via Boost escribió:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I am requesting to merge a small bugfix to master on Boost.Math (https://github.com/boostorg/math/pull/1659). iround has a forward declaration marked with BOOST_MATH_GPU_ENABLED, but the function definition did not. Previously, this annotation was implicitly added in CUDA, but that no longer is the case so an error occurs when iround is used in CUDA kernels.
All checks seem ok, it's a very small bugfix. Go ahead! Ion
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Log develop to master: https://github.com/boostorg/log/compare/master...develop There are quite a few bug fixes, mostly related to asynchronous_sink: * Added missing memory order constraints to threadsafe_queue methods, which should fix thread synchronization for asynchronous_sink using unbounded_fifo_queue. * Added missing memory order constraints to the once_block utility, which is used in multiple places throughout the library to implement one-time initialization. This should fix potential issues, when a thread might not observe the effects of initialization performed in another thread, after it passes the once block on the fast path. * Fixed a bug, where a flush of asynchronous_sink with a blocking queue could result in dropping log records waiting for free room in the queue. * Fixed an unlikely bug on Windows, where interprocess_condition_variable::wait (which is used in IPC queue implementation) could have used an already freed semaphore handle. If the handle was reused by the OS, the implementation could have blocked for unknown amount of time. * Use atomic to store the pointer to NtQuerySemaphore function on Windows. Mostly to make TSan and similar tools happy. * Fixed a bug in asynchronous_sink::flush, where it could have not flushed non-ready log records in an ordering queue. The records would remain in the queue and processed later normally, provided that log record processing continues. * Fixed that asynchronous_sink::flush and consume could remain blocked if log record processing failed with an exception during the flush. * Updates to docs, tests and CI.
El 21/11/2026 a las 1:48, Andrey Semashev via Boost escribió:
On 12 Nov 2026 03:36, Marshall Clow via Boost wrote:
Release open for bug fixes and documentation updates. Other changes by permission of a release manager.
I would like to merge Boost.Log develop to master:
Please go ahead, quite a few changes, let's merge ASAP and make sure master regression tests are ok. Best, Ion
participants (5)
-
Andrey Semashev -
Emil Dotchevski -
Ion Gaztañaga -
Marshall Clow -
Matt Borland