Repository navigation
napari-virtual-reader v0.0.1
napari-virtual-reader v0.0.1
Status: Release prepared; GitHub Actions and PyPI publication pending.
Release preparation date: 2026-11-05.
Highlights
- Initial experimental Alpha release of napari-virtual-reader.
- Open supported images as lazy, reference-backed OME-Zarr without pixel conversion.
- Use napari's File/Open reader or the asynchronous plugin widget.
What Changed
- Registered the reader and widget through the npe2
napari.manifestentry point. - Added per-channel layers with OME axes, transforms, pyramids, colors, and display windows.
- Added compatible CZI scene stacking while retaining original scene metadata.
- Added atomic VZIP reference-archive saving with overwrite confirmation, without copying pixels.
- Added a shared, read-only inspector for Zarr structure and selected-layer metadata.
- Renamed napari-refkv to napari-virtual-reader, including its Python module and repository.
- Added automated tests and a test-gated, tag-triggered PyPI publishing workflow.
Validation
- Local Windows test suite: 32 tests passed.
- Ruff lint checks passed.
- Source distribution and wheel built successfully with Hatchling.
- Strict Twine metadata and README checks passed for both distributions.
- npe2 discovered a non-editable wheel installation and resolved its reader and widget commands.
- Public PyPI has compatible releases for the declared runtime dependencies.
- GitHub Actions tests Windows, Linux, and macOS with Python 3.12-3.14; release-run results are pending.
Publication
- Release tag:
v0.0.1(pending creation). - Release commit: The commit referenced by the annotated
v0.0.1tag. - GitHub release: v0.0.1.
- PyPI version: 0.0.1.
The URLs above are publication targets, not confirmation of successful publication.
Known Limitations
- Experimental Alpha software: APIs and behavior may change without notice.
- Python 3.12 or newer, napari 0.10.0 or newer, and a working Qt backend are required.
- Original sources must remain available and unchanged while layers or VZIP archives reference them.
- Format support depends on refkv and the source's metadata and codecs, not just its extension.
- Only local images and VZIP archives with supported OME metadata are accepted.
- Remote inputs and non-image datasets are outside this release's scope.
- Incompatible CZI scenes remain separate rather than being cropped or resampled.
- Missing display windows use dtype bounds; floating-point images default to 0-1 contrast.
- Contrast may need manual adjustment.
- Validate results independently before relying on them for analysis.