Skip to content

chore(rate-limiter): bump cpex_rate_limiter to 0.0.4 - #47

Merged
brian-hussey merged 1 commit into
mainfrom
bump/rate-limiter-v0.0.4
Apr 22, 2026
Merged

brian-hussey merged 1 commit into
mainfrom
bump/rate-limiter-v0.0.4

Conversation

@gandhipratik203

@gandhipratik203 gandhipratik203 commented Apr 22, 2026

Copy link
Copy Markdown
Collaborator

Summary

Bumps cpex_rate_limiter from 0.0.3 → 0.0.4 so the production-hardening fixes in #40 can be consumed downstream. PyPI 0.0.3 is already published and immutable, so a version bump is required for the main repo (mcp-context-forge) to pick up the fixes.

CI sentinel rotation

release-validation / resolve in the called release workflow enforces tag version == Cargo/manifest version (via tools/plugin_catalog.py::release_info). The PR-triggered smoke test was using rate-limiter-v0.0.3 as its sentinel, which now mismatches because this PR raises rate_limiter to 0.0.4.

This PR rotates the sentinel in ci-rust-python-package.yaml and ci-install-built-wheel.yaml off the plugin being bumped, onto pii-filter-v0.2.1 (the most recently published tag whose Cargo on main still matches). This keeps release-validation meaningful (it still replays a real released tag) without coupling it to the plugin currently being bumped. Every future bump of a sentinel plugin will need the same rotation.

Files touched

  • plugins/rust/python-package/rate_limiter/Cargo.toml — 0.0.3 → 0.0.4
  • plugins/rust/python-package/rate_limiter/cpex_rate_limiter/plugin-manifest.yaml — 0.0.3 → 0.0.4
  • Cargo.lock — auto-updated
  • .github/workflows/ci-rust-python-package.yaml — sentinel → pii-filter-v0.2.1
  • .github/workflows/ci-install-built-wheel.yaml — sentinel → pii-filter-v0.2.1
  • tests/test_plugin_catalog.py — 2 release-info tests updated to rate-limiter-v0.0.4, generated-workflow assertion updated to pii-filter-v0.2.1

Local verification

python3 -m unittest tests.test_plugin_catalog → 71 passed

Post-merge

Tag main HEAD as rate-limiter-v0.0.4; the tag push triggers the release workflow and publishes the wheel to PyPI. The main repo (mcp-context-forge) can then pin cpex-rate-limiter==0.0.4 to consume #40's fixes.

Bumps the declared version of cpex_rate_limiter from 0.0.3 to 0.0.4.

Version-bearing files (coupled via tools/plugin_catalog.py release_info
enforcement that tag version == Cargo/manifest version):

  - plugins/rust/python-package/rate_limiter/Cargo.toml
  - plugins/rust/python-package/rate_limiter/cpex_rate_limiter/plugin-manifest.yaml
  - Cargo.lock                                          (auto-updated)
  - tests/test_plugin_catalog.py                        (2 release-info
    tests that validate the rate_limiter canonical tag)

Because release-validation / resolve validates the plugin referenced by
the ci-yaml `tag:` input (via release_info reading the PR-head Cargo),
the bump of rate_limiter breaks the prior sentinel. Rotate both PR-
triggered release-validation calls off rate-limiter onto a stable
already-released sentinel. retry-with-backoff-v0.1.1 was chosen because
(a) its Cargo on main matches the tag version, and (b) it has not been
used as a PR-smoke sentinel before, so the test.pypi publish path
produces a fresh upload rather than colliding on content hash with a
prior repeated sentinel upload:

  - .github/workflows/ci-rust-python-package.yaml       (tag: retry-with-backoff-v0.1.1)
  - .github/workflows/ci-install-built-wheel.yaml       (tag: retry-with-backoff-v0.1.1)
  - tests/test_plugin_catalog.py                        (generated-
    workflow assertion updated to match)

Naming-example references in DEVELOPING.md, README.md, and AGENTS.md
continue to use rate-limiter-v0.0.2 as their illustrative placeholder
and are intentionally not touched.

Local gate:
  python3 -m unittest tests.test_plugin_catalog → 71 passed

Signed-off-by: Pratik Gandhi <gandhipratik203@gmail.com>
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.

3 participants