Up to [cvs.NetBSD.org] / pkgsrc / parallel / py-threadpoolctl
Request diff between arbitrary revisions
Keyword substitution: kv
Default branch: MAIN
py-threadpoolctl: updated to 3.8.0 3.8.0 Run faster on Linux, where the new /proc/self/maps mechanism in 3.7.0 added quite a bit of overhead. When OpenBLAS uses a non-OpenMP threading layer, you can only set the size of its thread pool from a single controlling thread; since it has a shared process-wide thread pool, having multiple Python threads fight over control wouldn't work well. A new API OpenBLASController.clear_controlling_thread() allows resetting which thread controlls OpenBLAS in these cases, and decorator and context manager limit APIs will automatically call that API when they are done. Add a new API to get a cached ThreadpoolController instance (get_cached_controller()), as an optimization for cases where limits are set frequently. See the docstring for caveats. Added official support for Python 3.15, and declared support for Python 3.14 in package metadata. Faster performance when creating a ThreadpoolController. Shared libraries are now cached across runs, whereas previously they would be unloaded after the ThreadpoolController was garbage collected. A single shared library can no longer have multiple LibController instances. When passing a list of dicts or a dict to threadpool_limits(limit=...) and related APIs, the "filepath" is used as the unique key identifying LibControllers, if available, falling back to "prefix". Previously the "prefix" was checked first. Fixed limits="sequential_blas_under_openmp", which left all BLAS libraries unlimited as soon as OpenBLAS was loaded, whatever its threading layer, or a library used the OpenMP threading layer (e.g. BLIS). The limit is now only skipped when OpenBLAS uses the OpenMP threading layer, as documented. Dropped official support for Python 3.9. Fixed recursive @threadpool_limits.wrap() calls so they restore limits correctly. Fixed @threadpool_limits.wrap() so it can use third-party libraries that were registered or imported after the function was decorated.
py-threadpoolctl: updated to 3.7.0 3.7.0 (2026-09-15) Fixed an intermittent OSError on Windows when DLLs are loaded or unloaded concurrently during library discovery (for example when importing conda-forge OpenCV). On Python 3.14+, discovery uses ctypes.util.dllist when available. If dllist raises OSError, threadpoolctl emits a RuntimeWarning instead of crashing so the failure can be reported upstream with a minimal reproducer. Older Pythons use a Toolhelp snapshot enumerator, with graceful per-module fallbacks. Added the ability to check whether a limiting API affects just the current thread or the whole process. Mainly aimed at debugging and diagnostics, and somewhat unreliable, it is therefore enabled by default only for command-line usage. Only warn about simultaneous libomp and libiomp usage on Linux, where the incompatibility is known to cause crashes. Fixed a deadlock triggered by getting or setting MKL's number of threads from parallel threads when using MKL with libiomp (Intel threading) on Linux. Going forward, setting the number of threads will only have a thread-local impact if feasible (for example, at minimum the underlying library must support this option, and many don't.) For MKL, setting the number of threads is now thread-local, i.e. limiting the number of threads won't impact MKL's thread pool size when using MKL in other Python threads. For OpenBLAS compiled with OpenMP on Linux and macOS, setting the number of threads is now thread-local, i.e. won't impact OpenBLAS thread pool size in other Python threads. On Windows behavior is likely process-wide, but this may depend on how OpenBLAS was compiled with OpenMP. Fix OpenBLAS detection for conda package on Windows Fixed a deadlock on Linux when using threadpoolctl from multiple threads. Start using Python 3.14's built-in support for listing shared libraries. On Linux, start using /proc/self/maps for listing shared libraries. Avoid importing ctypes.util on Linux (and load libc with ctypes.CDLL(None)) so threadpool_info() does not create libffi closures that can abort after os.fork() on some libffi builds. Dropped official support for Python 3.9.
py-threadpoolctl: fix build with flit_core 4.
py-threadpoolctl: updated to 3.6.0 3.6.0 Added support for libraries with a path longer than 260 on Windows. The supported path length is now 10 times higher but not unlimited for security reasons. Dropped official support for Python 3.8.
py-threadpoolctl: adapt for flit_core 3.11. Bump PKGREVISION.
py-threadpoolctl: updated to 3.5.0 3.5.0 (2024-04-29) - Added support for the Scientific Python version of OpenBLAS (https://github.com/MacPython/openblas-libs), which exposes symbols with different names than the ones of the original OpenBLAS library. https://github.com/joblib/threadpoolctl/pull/175
py-threadpoolctl: updated to 3.4.0 3.4.0 (2024-03-20) ================== - Added support for Python interpreters statically linked against libc or linked against alternative implementations of libc like musl (on Alpine Linux for instance). https://github.com/joblib/threadpoolctl/pull/171 - Added support for Pyodide https://github.com/joblib/threadpoolctl/pull/169
py-threadpoolctl: updated to 3.3.0 3.3.0 (2024-02-14) ================== - Extended FlexiBLAS support to be able to switch backend at runtime. https://github.com/joblib/threadpoolctl/pull/163 - Added support for FlexiBLAS https://github.com/joblib/threadpoolctl/pull/156 - Fixed a bug where an unsupported library would be detected because it shares a common prefix with one of the supported libraries. Now the symbols are also checked to identify the supported libraries. https://github.com/joblib/threadpoolctl/pull/151
python/wheel.mk: simplify a lot, and switch to 'installer' for installation This follows the recommended bootstrap method (flit_core, build, installer). However, installer installs different files than pip, so update PLISTs for all packages using wheel.mk and bump their PKGREVISIONs.
py-threadpoolctl: updated to 3.2.0 3.2.0 (2023-07-13) ================== - Dropped support for Python 3.6 and 3.7. - Added support for custom library controllers. Custom controllers must inherit from the `threadpoolctl.LibController` class and be registered to threadpoolctl using the `threadpoolctl.register` function. https://github.com/joblib/threadpoolctl/pull/138 - A warning is raised on macOS when threadpoolctl finds both Intel OpenMP and LLVM OpenMP runtimes loaded simultaneously by the same Python program. See details and workarounds at https://github.com/joblib/threadpoolctl/blob/master/multiple_openmp.md. https://github.com/joblib/threadpoolctl/pull/142
py-threadpoolctl: added version 3.1.0 Python helpers to limit the number of threads used in the threadpool-backed of common native libraries used for scientific computing and data science (e.g. BLAS and OpenMP).