Skip to content

chore: hatchling/hatch-vcs dynamic versioning, release CI, Makefile, and dev docs#16

Open
H00N24 wants to merge 8 commits into
masterfrom
ondrej/uv-and-pypi-release
Open

chore: hatchling/hatch-vcs dynamic versioning, release CI, Makefile, and dev docs#16
H00N24 wants to merge 8 commits into
masterfrom
ondrej/uv-and-pypi-release

Conversation

@H00N24
Copy link
Copy Markdown

@H00N24 H00N24 commented May 5, 2026

Summary

Consolidate graphcore's CI into a single workflow and adopt uv-native publishing
to the Certora private PyPI (AWS CodeArtifact).

Changes

  • Single CI workflow (.github/workflows/ci.yml) replacing pyright.yml,
    pytest.yml, and release.yml. Flow: build_package → tests → codeartifact_pypi_upload.
    • tests downloads the wheel artifact and installs it (uv pip install --force-reinstall --no-deps dist/*.whl),
      so CI exercises the built package, not editable source.
    • pytest and pyright run as steps in the tests job with if: success() || failure()
      so a failure in one doesn't skip the other.
    • Publish job runs only on push to main/master or v*.*.* tags, gated on tests passing.
  • uv publish --index certora replaces uvx twine upload and the
    aws codeartifact get-repository-endpoint lookup. The publish/check URLs
    come from a new [[tool.uv.index]] block in pyproject.toml (with
    explicit = true, so public PyPI stays the default for resolution).
  • Action pinning: every action pinned to a commit SHA with a trailing
    # vX.Y.Z comment.
  • Dependabot (.github/dependabot.yml) added — weekly grouped bumps for
    GitHub Actions (keeps the SHA pins fresh).
  • Permissions tightened to a global contents: read, with per-job grants
    (id-token: write for AWS OIDC, contents: write only for the release job).
  • Makefile: added pytest and pyright targets running through uv run,
    so CI and local dev share the same entrypoints.
  • Removed requirements.txt (duplicated uv.lock).

Potential future plans

  • Adding Certora prefix to the package

More information about prefix etc https://github.com/Certora/certora-python-cookiecutter

H00N24 and others added 3 commits May 5, 2026 14:46
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
@H00N24 H00N24 self-assigned this May 5, 2026
@H00N24 H00N24 changed the title [WIP] chore: hatchling/hatch-vcs dynamic versioning, release CI, Makefile, and dev docs chore: hatchling/hatch-vcs dynamic versioning, release CI, Makefile, and dev docs May 7, 2026
@H00N24 H00N24 marked this pull request as ready for review May 7, 2026 09:02
H00N24 and others added 2 commits May 7, 2026 18:37
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
langgraph-prebuilt 1.0.13 imports `ExecutionInfo` from
`langgraph.runtime`, a symbol introduced after 1.0.5. The previous
strict pin caused an ImportError at test-collection time. Resolved
to 1.1.10.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@yoav-el-certora yoav-el-certora requested a review from jtoman May 11, 2026 06:02
Comment thread pyproject.toml Outdated
Comment on lines +20 to +18
"langgraph==1.0.5",
"langgraph>=1.0.13",
Copy link
Copy Markdown
Author

@H00N24 H00N24 May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This was updated to make the CI pass (langchain-ai/langgraph#7404) , do we want to lock to 1.0.5 with lock on langgraph-prebuilt?

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

upstream ai composer just pins to langgraph-prebuilt==1.0.8. More to the point, updating past 1.0.5 introduces a change in how langgraph handles checkpoints that is very disruptive for our tooling and extremely difficult for us to program around. In other words 1.0.5 is the max langgraph version we're going to be using for the time being.

Comment thread pyproject.toml Outdated
Comment on lines +20 to +18
"langgraph==1.0.5",
"langgraph>=1.0.13",
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

upstream ai composer just pins to langgraph-prebuilt==1.0.8. More to the point, updating past 1.0.5 introduces a change in how langgraph handles checkpoints that is very disruptive for our tooling and extremely difficult for us to program around. In other words 1.0.5 is the max langgraph version we're going to be using for the time being.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants