stillwater-sc / universal / 36099390330
86%
master: 84%

LAST BUILD BRANCH: main
DEFAULT BRANCH: master
Ran
Jobs 1
Files 676
Run time 1min
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
coverage: 85.99% (-0.002%) from 85.992%
36099390330

push

github

web-flow 
ci: remove the CodeQL workflow (#1606)

CodeQL has not produced a finding since 2026-05-02, nearly five months. Its
whole output was 37 alerts, all from 2026-03 and 2026-05, and every one is
already resolved: 34 fixed, 3 dismissed, 0 open. It was front-loaded value that
has since stopped arriving.

What it cost while producing nothing: ~35 minutes per run, 4-8 runs a day on
pushes to main, and 6.39GB of Actions cache across 31 per-commit TRAP entries --
62% of everything the repo had stored. It was already tuned down once, when the
pull_request trigger was removed for adding >20 minutes to every PR on a
template-heavy codebase. This is the same complaint, escalated.

The 6.39GB mattered more than its share suggests. The Actions cache went READ
ONLY on 2026-09-22 against the spend budget, not the 20GB eviction ceiling, and
every ccache save since has failed with "Cache reservation failed: You have
reached your configured budget". Restores kept working, so ccache hits froze at
the 296 objects the 2026-09-22 generation happened to hold while the denominator
grew with each new test -- 62% decaying to 57% on main, and 41%/22% on a branch
touching widely-included headers. That is the whole of the fast tier going from
~6 to ~16 minutes. The 31 TRAP caches have been deleted, taking stored usage
from 10.29GB to 3.80GB.

Deleting the workflow disables scanning completely: code-scanning default setup
reports state=not-configured, so this file was the only thing running it. CodeQL
is not in the branch protection required set (check-ascii, build, lint-pr-title),
so nothing is blocked. No README badge referenced it.

Restoring it later is a revert of this commit. If the weekly drift scan turns out
to be worth keeping without the per-push cost, the cheaper shape is this file
with the `push:` trigger dropped, leaving only the Wednesday cron -- roughly one
run a week instead of thirty.

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

45513 of 52928 relevant lines covered (85.99%)

9412790.74 hits per line

Coverage Regressions

Lines Coverage ∆ File
1
95.86
-0.06% include/sw/universal/number/cfloat/cfloat_impl.hpp
Jobs
ID Job ID Ran Files Coverage
1 36099390330.1 676
85.99
GitHub Action Run
Source Files on build 36099390330