Bump nanobind from 2.10.2 to 2.15.0

[dependabot]

Bumps nanobind from 2.10.2 to 2.15.0.

Changelog

Sourced from nanobind's changelog.

Version 2.15.0 (Aug 15, 2026)

Most changes in this release revolve around stub generation.

• The stubgen machinery for generating import statements was rewritten to fix brittleness and prevent it from generating import declarations that were invalid Python syntax. It now records whether a module is imported out of convenience or because it should be re-exported, and it writes self-imports in absolute form. Standard library modules are identified via sys.stdlib_module_names, which changes how imports are grouped and ordered. Parsing and escaping of type and docstring text was made more robust in several places. (commits 8df192 <https://github.com/wjakob/nanobind/commit/8df1920f8cd8f4a7e2dae509fe778a05988e4ad7>, ab446f <https://github.com/wjakob/nanobind/commit/ab446f3c058c40013377caf27b5db8eeb98bff88>, 39af7b <https://github.com/wjakob/nanobind/commit/39af7ba928f062847ace31766f52b8fe8b7514db>, 9a7d3d <https://github.com/wjakob/nanobind/commit/9a7d3d79ab917f286136cc0de6dd315db1b05373>).

• Stubs for :cpp:class:nb::ndarray <ndarray> arguments besides NumPy are now valid Python. They previously carried extra information in square brackets, as in mlx.core.array[dtype=float32, shape=(2, 4)], which tools like ruff and ty reject. (PR [#1409](https://github.com/wjakob/nanobind/issues/1409) <https://github.com/wjakob/nanobind/pull/1409>__).

• stubgen now accepts multiple pattern files via -p, and the PATTERN_FILE parameter of the CMake :cmake:command:nanobind_add_stub command accepts a list. (commit 581adb <https://github.com/wjakob/nanobind/commit/581adb8cf70ed38d4dbbeb9969113ea36fac084a>__).

• When a pattern file modifies an overload chain with one \doc marker per overload, each marker now expands to the docstring of the associated overload. (commit 5031de <https://github.com/wjakob/nanobind/commit/5031dee02fdffff7486d31cb393e051d39421c50>__).

• Annotations that name a module through an alias now resolve to the real module name. For example, f(x: np.ndarray) is rendered as f(x: numpy.ndarray). When processing files with from __future__ import annotations, stubgen previously generated incorrect output like import np. (commit 23051a <https://github.com/wjakob/nanobind/commit/23051ac3420b8ce9c009843ecc1eea976ee9fecd>__).

• Properties with a setter but no getter are rendered as a plain annotated attribute. Previous versions emitted a @property decorator without an accompanying function definition. (commit ab446f <https://github.com/wjakob/nanobind/commit/ab446f3c058c40013377caf27b5db8eeb98bff88>__).

• stubgen now sets the environment variable NB_STUBGEN to "1". Extensions can query this variable to skip initialization steps that are expensive or inappropriate for stub generation (see the section on :ref:detecting stub generation <stubgen_detection> for details). (commit ee6516 <https://github.com/wjakob/nanobind/commit/ee6516d07735ee1e7f1386f7d567ece6248d7e9c>__).

... (truncated)

Commits

• a479830 v2.15.0 release
• 463392a changelog updates
• b8a4446 Change type name to tensorflow.Tensor in nb::ndarray
• 0c0a5fb Put dtype in quotes for non-numpy array annotations
• b2d3bc1 Add pytorch to array framework lookup table
• 01bb706 Correctly import nested array modules, process annotations like for NumPy
• ff85816 docs: Fix minor typos
• ee6516d stubgen: set the NB_STUBGEN environment variable
• 16eef7f changelog update
• 53d546f stubgen: unify function dispatch in put()
• Additional commits viewable in compare view