Metadata-Version: 2.4
Name: dcccpy-windows-runtime
Version: 0.1.0
Summary: Windows DCCCcore runtime package for dcccpy
Author: DCCCSlicer contributors
License-Expression: CC-BY-NC-ND-4.0
Project-URL: Homepage, https://github.com/tctco/DCCCSlicer
Project-URL: Repository, https://github.com/tctco/DCCCSlicer
Project-URL: Issues, https://github.com/tctco/DCCCSlicer/issues
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Science/Research
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3 :: Only
Classifier: Operating System :: Microsoft :: Windows
Classifier: Topic :: Scientific/Engineering :: Medical Science Apps.
Requires-Python: >=3.9
Description-Content-Type: text/markdown

# dcccpy-windows-runtime

`dcccpy-windows-runtime` carries the prebuilt Windows `DCCCcore` runtime used by
`dcccpy[windows-runtime]` and by `dcccpy[runtime]` on Windows.

This package is optional. The plain `dcccpy` package can download `DCCCcore`
from GitHub releases on first use. Install this runtime package when a user or
environment needs `pip install` to provide the native runtime without first-run
download.

Before building this package, vendor the release asset:

```bash
python scripts/vendor_dccccore.py --version 4.2.3 --release-platform windows-latest-x64
python -m build --wheel
```

For a PyPI-size runtime wheel, use the slim profile:

```bash
python scripts/vendor_dccccore.py --profile pypi-slim --force
python -m build --wheel
```

The `pypi-slim` profile omits the `fast_and_acc` registration model/config and
the ADAD decoupler ONNX ensemble. It keeps the default spatial normalization
model and the assets needed by common Centiloid/CenTauR/CenTauRz workflows.

The vendored runtime is intentionally ignored by git because it contains large
ONNX model and NIfTI template assets.
