Files
worktrunk/.github/workflows/coverage.yaml
Worktrunk Botandworktrunk-bot 224e408b80 ci: bump cargo-nextest pin to 0.9.149 (#4586)
Bump the pinned cargo-nextest from 0.9.148 to 0.9.149 in the test and
Tend setup actions, coverage workflow, and Codex Cloud setup. The four
copies now agree. Version 0.9.149 updates the certificate trust used by
`cargo nextest self update`; the release says ordinary test runs do not
use that network path. Its crate declares Rust 1.91, below this
repository's Rust 1.98 toolchain.

Verification: `git diff --check` passed. This pin change needs the PR's
CI to exercise test and coverage setup.

[Nextest 0.9.149 release
notes](https://github.com/nextest-rs/nextest/releases/tag/cargo-nextest-0.9.149)

Co-authored-by: worktrunk-bot <254187624+worktrunk-bot@users.noreply.github.com>
2026-10-25 21:47:36 -07:00

87 lines
3.0 KiB
YAML

name: coverage
on:
push:
branches: [ main ]
pull_request:
branches: [ main ]
workflow_dispatch:
# Keyed per-commit on main, so a merge train can't cancel a queued upload:
# GitHub cancels the pending run whenever a newer one joins the group, and
# codecov picks a PR's base by walking back to the newest ancestor that has a
# report. A main commit that skips its upload therefore leaves later PRs
# comparing against a stale base. That's why coverage sits outside `ci`, whose
# group is shared with jobs that hold a run open for hours. PR runs still
# supersede their own.
concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}
cancel-in-progress: ${{ github.event_name == 'pull_request' }}
env:
# Swatinem/rust-cache hashes CARGO*/RUST* into the cache key, so these track
# ci.yaml (.github/AGENTS.md → Build environment).
CARGO_TERM_COLOR: always
CARGO_INCREMENTAL: 0
RUSTFLAGS: -C debuginfo=0
jobs:
code-coverage:
runs-on: ubuntu-24.04
steps:
- name: 📂 Checkout code
uses: actions/checkout@v7
- uses: baptiste0928/cargo-install@v3
with:
crate: cargo-llvm-cov
version: "=0.9.1"
- uses: baptiste0928/cargo-install@v3
with:
crate: cargo-nextest
version: "=0.9.149"
- name: 💰 Cache
uses: Swatinem/rust-cache@v2
with:
# Own cache — coverage-instrumented builds differ from the shared test
# cache. Already main-gated; cache-bin:false matches everywhere else.
cache-bin: "false"
save-if: ${{ github.ref == 'refs/heads/main' }}
- name: Install shells (zsh, fish)
run: sudo apt-get update && sudo apt-get install -y zsh fish
- name: Install nushell
uses: hustcer/setup-nu@v3
with:
version: '0.116.1'
# Ensure nothing remains from caching
- run: cargo llvm-cov clean --workspace
- name: 📊 Generate coverage report
# Match the regular suite's process isolation. Cargo's in-process runner
# lets one busy PTY test delay another before it can paint its prompt.
run: cargo llvm-cov nextest --features shell-integration-tests --cobertura --output-path=cobertura.xml
- name: Upload code coverage results
uses: actions/upload-artifact@v7
with:
name: code-coverage-report
path: cobertura.xml
- name: Upload to codecov.io
uses: codecov/codecov-action@v7.1.1
with:
use_pypi: true
version: 11.3.1
files: cobertura.xml
# Soft-fail on fork PRs: secrets aren't available there, so we rely on
# tokenless upload, which can hiccup. Keep hard-fail on the main repo.
# The head repo is what tells the two apart — `github.repository_owner`
# is the base repo's owner on a fork PR too.
fail_ci_if_error: ${{ github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository }}
token: ${{ secrets.CODECOV_TOKEN }}