18
0
mirror of https://github.com/MaSzyna-EU07/maszyna.git synced 2026-11-19 22:44:59 +01:00

Merge pull request #325 from MaSzyna-EU07/feat/ci-unit-test-coverage

CI: coverage of the core library by the unit tests
This commit is contained in:
maj00r
2026-11-11 22:25:27 +01:00
committed by GitHub
2 changed files with 137 additions and 0 deletions

135
.github/workflows/coverage.yml vendored Normal file
View File

@@ -0,0 +1,135 @@
# .github/workflows/coverage.yml
# UNTRUSTED CONTEXT. This workflow runs fork-authored code.
# It must never reference `secrets.*`. It only builds, runs the unit tests and
# uploads a report; nothing here feeds the SonarQube Cloud analysis.
#
# Coverage of the game's core library (eu07_core) by the unit tests, measured with
# gcov on a GCC build: the report goes to the summary of the run and to an artifact.
# It is not sent to SonarQube Cloud on purpose: the project's quality gate ("Sonar
# way") fails a pull request whose new code is covered in less than 80%, a condition
# that isn't evaluated as long as sonar-project.properties excludes every file from
# coverage. Wiring this report in would turn that condition on.
name: Coverage
on:
push:
branches:
- master
pull_request:
types: [ opened, reopened, synchronize ]
permissions: {}
# A newer push to a pull request cancels the run for its older commit; a push to master
# gets a group of its own (its commit), so the runs on master are never cancelled.
concurrency:
group: ${{ github.workflow }}-${{ github.event_name == 'pull_request' && format('pr-{0}', github.event.pull_request.number) || github.sha }}
cancel-in-progress: true
jobs:
coverage:
name: Unit test coverage (GCC)
runs-on: ubuntu-24.04
timeout-minutes: 60
permissions:
contents: read
steps:
- uses: actions/checkout@d23441a48e516b6c34aea4fa41551a30e30af803 # v6.1.0
with:
submodules: recursive
- name: Set up Python 3.14
uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1 # v6.3.0
with:
python-version: '3.14'
# the same packages as the Build workflow: the game is configured as a whole,
# though only the unit tests and the core library they link are built
- name: Install build dependencies
run: |
sudo apt-get update -y
sudo apt-get install -y \
build-essential \
libgl1-mesa-dev libglew-dev libglfw3-dev \
libpng-dev libopenal-dev libluajit-5.1-dev \
libserialport-dev libsndfile1-dev \
xorg-dev libxkbcommon-dev libwayland-dev wayland-protocols
python3 -m pip install --no-cache-dir --only-binary :all: gcovr==8.3
- name: Build current OpenAL Soft
run: |
git clone --depth 1 --branch 1.24.2 https://github.com/kcat/openal-soft.git /tmp/openal-soft
cmake -S /tmp/openal-soft -B /tmp/openal-soft/build \
-DCMAKE_BUILD_TYPE=Release -DALSOFT_EXAMPLES=OFF -DALSOFT_UTILS=OFF
cmake --build /tmp/openal-soft/build -j"$(nproc)"
sudo cmake --install /tmp/openal-soft/build
sudo ldconfig
# see the Build workflow
- name: Patch GLM for GCC
run: |
sed -i 's/GLM_FUNC_QUALIFIER std::string format(const char\* message, \.\.\.)/inline std::string format(const char* message, ...)/' \
ref/glm/glm/gtx/string_cast.inl
grep -n "std::string format(const char\* message" ref/glm/glm/gtx/string_cast.inl
# unoptimised, so every line keeps its own code and the counts are those of the source
- name: Configure (CMake)
run: |
cmake -S . -B build \
-DCMAKE_BUILD_TYPE=Debug \
-DWITH_BETTER_RENDERER=OFF \
-DWITH_ZMQ=OFF -DWITH_OPENVR=OFF -DUSE_LTO=OFF \
-DCMAKE_CXX_FLAGS="--coverage -O0 -Wno-attributes -fno-inline-functions-called-once" \
-DPython3_EXECUTABLE="$(which python3)"
# the tests alone: the game itself isn't linked, so they aren't run by its build
- name: Build the unit tests
run: cmake --build build --target eu07_unit_tests -j"$(nproc)"
- name: Run the unit tests
run: ./build/tests/unit/eu07_unit_tests
# the game's own sources only: no submodules, no vendored code, not the tests themselves.
# betterRenderer/ is left out for its thirdparty/yaml-cpp, which every target of the game links and so builds here
- name: Coverage report
run: |
# a failure of gcovr fails the step, though its output goes through tee
set -o pipefail
mkdir -p coverage-report/html
gcovr --root . build \
--exclude 'ref/' --exclude 'imgui/' --exclude 'stb/' --exclude 'betterRenderer/' \
--exclude 'tests/' --exclude 'build/' \
--gcov-ignore-parse-errors=negative_hits.warn_once_per_file \
--gcov-ignore-errors=no_working_dir_found \
--print-summary \
--txt coverage-report/coverage.txt \
--html-details coverage-report/html/index.html \
--sonarqube coverage-report/coverage-sonarqube.xml \
--json-summary-pretty --json-summary coverage-report/coverage-summary.json \
| tee coverage-report/summary.txt
{
echo '## Unit test coverage of the core library'
echo
echo 'GCC, unoptimised build, `eu07_unit_tests`. Lines run by the tests, per source file of the game the tests compile.'
echo
echo '```'
grep -E '^(lines|functions|branches):' coverage-report/summary.txt
echo '```'
echo
echo '<details><summary>Per file</summary>'
echo
echo '```'
cat coverage-report/coverage.txt
echo '```'
echo
echo '</details>'
} >> "$GITHUB_STEP_SUMMARY"
- name: Upload the coverage report
uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
with:
name: unit-test-coverage
path: coverage-report
retention-days: 14
if-no-files-found: error

View File

@@ -21,6 +21,8 @@ Coverage was measured on Windows with MSVC's SanitizerCoverage (`/fsanitize-cove
The aim is branch coverage with the rest explained below, not 100%.
The numbers above are from 2026-11-01. Since then the **Coverage** workflow (`.github/workflows/coverage.yml`) measures every pull request and every push to master: a GCC build of the unit tests with gcov, lines, functions and branches per source file of the core library. The totals and the per-file table are in the summary of the workflow run, and the HTML report with the lines marked is in its `unit-test-coverage` artifact. Its counts are source lines and gcov's branches, so they differ somewhat from the blocks in the table. The report isn't sent to SonarQube Cloud: the project's quality gate would then require 80% coverage of the new code of every pull request.
## Code nothing in the game runs
Nothing creates these objects or calls these functions anywhere in the game, which was checked against the whole source tree. A refactor can delete them, or leave them untested.