Version 0.59.1 (18 March 2024)

This is a bug-fix release to fix regressions in 0.59.0.

CUDA API Changes

Fixed caching of kernels that use target-specific overloads

Caching of kernels using target-specific overloads now works. This includes use of cooperative group sync, which is now implemented with a target-specific overload.

(PR-#9447)

Performance Improvements and Changes

Improvement to `np.searchsorted`

Fixed a performance regression introduced in Numba 0.59 which made np.searchsorted considerably slower.

(PR-#9448)

Bug Fixes

Fix issues with `np.searchsorted` not handling `np.datetime64`

This patch fixes two issues with np.searchsorted. First, a regression is fixed in the support of np.datetime64. Second, adopt NAT-aware comparisons to fix mishandling of NAT value.

(PR-#9445)

Allow use of Python 3.12 PEP-695 type parameter syntax

A patch is added to properly parse the PEP 695 syntax. While Numba does not yet take advantage of type parameters, it will no longer erroneously reject functions defined with the new Python 3.12 syntax.

(PR-#9459)

Pull-Requests:

PR #9445: Fix #9427 np.searchsorted on datetime64 (sklam)
PR #9447: Fix Issue #9432, caching of kernels using target-specific overloads (such as CG sync) (gmarkall sklam)
PR #9448: Fix np.searchsorted regression (guilhermeleobas)
PR #9449: Remove deprecated CondaEnvironment@1 (sklam)
PR #9450: Fix gpuci versions (gmarkall)
PR #9459: Support pep695 type param syntax (sklam)
PR #9491: Fix non-deterministic bug caused by unstableness in SSA (sklam, loicdtx)