chore: Add official Python 3.15 support

.github/workflows/checks.yml

@@ -29,6 +29,10 @@ jobs:
           - python-version: '3.12'
           - python-version: '3.13'
           - python-version: '3.14'  # To actually run the default config
+          # cvxpy is excluded on Python 3.15 because one of its dependencies (sparsediffpy<0.7) has no
+          # cp315 wheel and can't be built without BLAS.
+          - python-version: '3.15'
+            options: 'quadprog_projector fairgrad'
           # OS variations
           - os: macOS-latest
           - os: windows-latest
@@ -46,11 +50,6 @@ jobs:
             extra_groups: 'lower_bounds'
           # Latest: torch nightly build.
           - torch_channel: 'nightly'
-          # Upcoming Python version, not yet supported by stable torch. cvxpy is excluded because one
-          # of its dependencies (sparsediffpy) has no wheel yet and can't be built without BLAS.
-          - python-version: '3.15'
-            torch_channel: 'nightly'
-            options: 'quadprog_projector fairgrad'
 
     steps:
       - name: Checkout repository

CHANGELOG.md

@@ -8,6 +8,10 @@ changelog does not include internal changes that do not affect the user.
 
 ## [Unreleased]
 
+### Added
+
+- Added Python 3.15 classifier in pyproject.toml (we now also run tests on Python 3.15 in the CI).
+
 ### Fixed
 
 - Fixed `Krum` and `KrumWeighting` ranking rows by sums of Euclidean distances instead of

docs/source/installation.md

@@ -5,7 +5,7 @@
 :end-before: <!-- end installation -->
 ```
 
-Note that `torchjd` requires Python 3.10, 3.11, 3.12, 3.13 or 3.14 and `torch>=2.0`.
+Note that `torchjd` requires Python 3.10, 3.11, 3.12, 3.13, 3.14 or 3.15 and `torch>=2.0`.
 
 Some aggregators have additional dependencies that are not included by default when installing
 `torchjd`. The following table lists the optional dependency groups and the aggregators they enable:
@@ -20,3 +20,7 @@ To install `torchjd` with all of its optional dependencies, you can also use:
 ```
 pip install "torchjd[full]"
 ```
+
+Note that on Python 3.15, [cvxpy](https://github.com/cvxpy/cvxpy/) doesn't provide wheels yet, so
+installing `cagrad`, `nash_mtl` or `full` requires building some of its dependencies from source
+(which needs a BLAS library).

pyproject.toml

@@ -28,6 +28,7 @@ classifiers = [
     "Programming Language :: Python :: 3.12",
     "Programming Language :: Python :: 3.13",
     "Programming Language :: Python :: 3.14",
+    "Programming Language :: Python :: 3.15",
     "Topic :: Scientific/Engineering",
     "Topic :: Scientific/Engineering :: Artificial Intelligence",
 ]