Skip to main content

k-dash

k-dash implements the frozen v1 design in docs/design/: content-addressed TVM-FFI CUDA kernel builds, direct OCI Distribution API access, an atomic local cache, Docker/Nix AOT builds, and synchronous Python loading.

The runnable example is examples/axpy. It exports an AXPY CUDA kernel as one TVM-FFI kernel.so and includes an AFS/H20 smoke test.

python -m pip install -e '.[test]'
pytest
k-dash --help

The public Python API is intentionally one function:

import k_dash

module = k_dash.load("examples/axpy", version="1", jit_args={"block_size": 256})

Source ignore patterns are relative to the project root. Directory patterns such as build, build/, and build/** exclude the whole subtree without walking it. Existing file globs remain supported; required source files cannot be excluded. Symbolic links are not packaged or traversed.

In an already isolated CI container without a Docker daemon, k-dash publish --backend nix ... builds AOT artifacts directly with installed Nix. This explicit backend disables Nix's nested sandbox and build-user group; the CI container must provide isolation. It uses the same frozen release source, locked dependencies, BuildSpec and binary validation as Docker AOT. Docker remains the default. Registry credentials are not forwarded to the Nix build subprocess environment.

Releasing to PyPI

Package versions come from git describe through hatch-vcs. An exact tag such as v0.1.0 builds version 0.1.0; commits after a tag receive a development version. Build from a Git checkout with its tags available.

Before the first release, configure a PyPI Trusted Publisher for project k-dash (use a pending publisher if the project does not yet exist):

  • GitHub owner: reyoung
  • Repository: k-dash
  • Workflow: publish.yml
  • Environment: pypi

The workflow uses OIDC, so no PyPI API token secret is required. Push a release commit and its annotated tag to build, validate, and publish the wheel:

git tag -a v0.1.0 -m "Release v0.1.0"
git push origin HEAD
git push origin v0.1.0

The workflow retains the built wheel as a GitHub Actions artifact and publishes it only after its metadata and version match the release tag.

Metadata

Release files for k-dash 0.1.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distribution (wheel)

Table of built distributions (wheels) for k-dash 0.1.2
File Interpreter ABI Platform
k_dash-0.1.2-py3-none-any.whl Python 3 none any Details

Release files / k_dash-0.1.2-py3-none-any.whl

Download URL k_dash-0.1.2-py3-none-any.whl
Size 39.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
ab254aad3833728d7110cd344fd521bab6bfae6dffb0cb8068d7e66a9ba8aaa4
BLAKE2b-256 checksum
How to use checksums
eb063e718f6015061f6c15628ccf0e8ca01e46f26e3be8ce9f6f06d2eff0a17a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 13, 2026.

Transparency log

Security for k-dash 0.1.2

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

All 1 file in this release has been uploaded with provenance metadata.

Attestation for k_dash-0.1.2-py3-none-any.whl

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 13, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page