[eclipse-score-bot]

This PR contains the following updates:

Package	Type	Update	Change
score_coverage	bazel_dep	patch	0.4.1 → 0.4.2

---

Release Notes

eclipse-score/coverage_tool (score_coverage)

v0.4.2

What's Changed

• scope: a cc_binary executable is the baseline object of its sources by @​dcalavrezo-qorix in #​44
• Migrate coverage summary checks from shell to pytest by @​AlexanderLanin in #​38
• docs: date the 0.4.2 release notes by @​dcalavrezo-qorix in #​45

Full Changelog: eclipse-score/coverage_tool@v0.4.1...v0.4.2

In plain words

The sources of a binary show up at 0 % like everything else. A
cc_binary links its own files, typically main.cpp, straight into the
executable instead of into a library archive. The reporter took its
zero-coverage baseline from archives only, so an untested binary's own files
were listed in unmapped_files.txt as no-data and were absent from the
report (lifecycle's launch_manager, #​40). The coverage-built executable now
serves as their baseline object, as it already did for rust_binary.

And the HTML tells the truth about them. With an executable among the
baseline objects, the single llvm-cov show run collided on main, the one
function name every test binary also has: with the same structural hash
llvm-cov attributed a test's counts to the untested file (100 %), with a
different hash it dropped the file from the index. Baseline-only files are
now rendered in their own --empty-profile run and spliced into the report,
the way the LCOV path already worked, and the index totals the
effective-coverage gate reads are recomputed. Validated on lifecycle:
main.cpp at 0 of 3 lines in HTML and LCOV alike.

Details for integrators

• Scope aspect: for a cc_binary in or below the scope roots, the executable
  joins the baseline objects; libraries are unchanged. Nothing to configure.
• gcov backend: the .gcno notes of untested files are built in the exec
  configuration, which takes the host options. Add
  --host_cxxopt=-O0 next to --cxxopt=-O0 in the gcov config (user manual
  step 4b), or GCC optimises those units and their line tables differ from
  the -O0 test builds; a folded return 0 then lacks its line. The LLVM
  backend is unaffected (front-end mapping).
• Reports of repositories with binaries in scope change: those files now
  count in the HTML totals and in the effective coverage, as they always did
  in the LCOV. Expect the effective figure to move down by the untested
  binary sources.
• Tests: 18 analysis tests, 74 reporter unit tests, 28 end-to-end checks on
  both backends (index rows of the untested sources at 0 %); the markdown
  summary checks moved from the shell script to pytest (#​38). Manual section
  "Binaries in the scope", verification report counts updated.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by #infrastructure automation. Contact us if you have any questions or feedback.

[github-actions]

Documentation preview for this pull request is available at:
pr-813: https://eclipse-score.github.io/lifecycle/pr-813/

[NicolasFussberger]

Will be done in #804

[eclipse-score-bot]

Renovate Ignore Notification

Because you closed this PR without merging, Renovate will ignore this update (0.4.2). You will get a PR once a newer version is released. To ignore this dependency forever, add it to the ignoreDeps array of your Renovate config.

If you accidentally closed this PR, or if you changed your mind: rename this PR to get a fresh replacement PR.