vbpf / prevail
86%
main: 86%

LAST BUILD BRANCH: ci/coverage-release-only
DEFAULT BRANCH: main
Repo Added
Token PQR6FR1EvpiPOMjKbwdwyvS0dofSolOGj regen
Build 962 Last
Files 79
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

LAST BUILD ON BRANCH ci/coverage-release-only
branch: ci/coverage-release-only
CHANGE BRANCH
x
Sync Branches

coverage: 85.986% (-0.9%) from 86.872%
30765435067

Pull #1220

github

elazarg 
Measure coverage on Release only and file it against the reviewed commit

Two problems, both visible in the runs from today.

The Debug leg cost between 56 min and 2h34m per run while the Release leg of the
same matrix took 8-10 min:

    main de799d87   Release  8 min   Debug     56 min
    PR #1216        Release  9 min   Debug     65 min
    PR #1217        Release 10 min   Debug  2h34m

An unoptimized build of the verifier, instrumented with --coverage
-fprofile-update=atomic and running the whole sample corpus, is that much slower
than an optimized one, and it gated every pull request. Debug is still compiled,
tested, and leak-checked by the CPP CI workflow, so no configuration stops being
exercised; only the second instrumented build goes away. With one leg left, the
parallel upload and its finish job have nothing to coordinate, so both go too.

Measuring Release alone may move the reported percentage slightly, since code
compiled out under NDEBUG (assertions) stops being counted as relevant lines
rather than as uncovered ones. The trend and the badge stay meaningful.

Second, git-branch and git-commit default to GITHUB_REF and GITHUB_SHA, which on
a pull_request event name the ephemeral merge ref, not the branch and commit
under review. Every pull request build was therefore filed against a commit that
exists on no branch -- Coveralls recorded ea68be35 ("Merge 4b26c5a6 into
de799d87") for #1218, whose head is 4b26c5a6. A status posted there lands on a
SHA the pull request never shows, and the coverage delta compares against a
phantom parent. Name the head explicitly; on push both expressions fall back to
the pushed ref.

Note that Coveralls comments additionally require the @coveralls user to have
write access to the repository and 'LEAVE COMMENTS?' enabled in the Coveralls
repo settings; neither is visible from CI.

Also drops a stale comment left over from an earlier debugging change.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthr... (continued)
Pull Request #1220: Measure coverage on Release only and file it against the reviewed commit

9026 of 10497 relevant lines covered (85.99%)

3151574.29 hits per line

Relevant lines Covered
10497 RELEVANT LINES 9026 COVERED LINES
3151574.29 HITS PER LINE
Source Files on ci/coverage-release-only

Recent builds

Builds Branch Commit Type Ran Committer Via Coverage
30765435067 ci/coverage-release-only Measure coverage on Release only and file it against the reviewed commit Two problems, both visible in the runs from today. The Debug leg cost between 56 min and 2h34m per run while the Release leg of the same matrix took 8-10 min: main de7... Pull #1220 elazarg github
85.99
See All Builds (939)

Badge your Repo: prevail

We detected this repo isn’t badged! Grab the embed code to the right, add it to your repo to show off your code coverage, and when the badge is live hit the refresh button to remove this message.
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

Refresh