Skip to content

feat(hydra-gates): SIG v18.1 star bands for gates 122 and 123, and gates.json per run - #909

Merged
rubenvdlinde merged 3 commits into
mainfrom
feat/sig-v18-star-bands-and-gates-json
Nov 6, 2026
Merged

rubenvdlinde merged 3 commits into
mainfrom
feat/sig-v18-star-bands-and-gates-json

Conversation

@rubenvdlinde

@rubenvdlinde rubenvdlinde commented Nov 5, 2026 •

Copy link
Copy Markdown
Contributor

What changes

Gate 122 (sig-maintainability) rates against the 4-star limits of the SIG/TÜV NORD CERT Guidance for Producers v18.1 (31 March 2026, the 2026 model): unit size 48.0/24.2/9.1%, complexity 20.7/7.5/1.1%, interfacing 15.4/3.4/0.9%, coupling 9.7/5.3/1.8%, duplication 5.8%, entanglement 0.075, independence 94.5%. Volume is not rated.

Every report row now carries stars, 5 to 1:

  • 5, 3 and 2 stars: the v18.1 limits scaled by the band-to-4-star ratios of Alves, Correia and Visser 2011 (Table III); duplication by the Heitlager 2007 band ratios. 1 star is anything worse than 2.
  • Independence and entanglement: 5 stars at half the 4-star exposure (97.3% / 0.0375), no band under 4 (<4). Entanglement is rated on our violation-degree estimate and is not a floored check.
  • The five-star profile uses the derived 5-star band (36.0/17.0/5.5% and so on) instead of half of each 4-star limit.

Gate 123 (sig-floor) records the guidance in the floor file. write never lowers a floor: a floored check the new limits miss stays in the file and is printed as "floored but missed at the current limits". A check that left the measured scope still leaves.

gates.json. The runner writes gates.json next to the findings logs ($HYDRA_GATE_GATES_JSON moves it): every declared gate with gate, name, verdict, findings, findings_from, scope, category and reason. Findings come from the count in the verdict reason, else from the gate's findings log, else null. A gate that emitted nothing is not-run. The writer cannot change the run's verdict.

Checks whose 4-star status flips on v18.1

All 20 apps, measured on development on 5 November. FLOORED means the check is in the app's floor. A floored check that flips to missed (filinq, integriq, opencatalogi independence; humaniq coupling >50) makes gate 123 fail on a run without a delta base until development meets it again. buildiq and shillinq vue unit_complexity >5 were already floored and missed under v17.

  • buildiq | no flips | floored+missed now: ['vue unit_complexity >5']
  • decidiq | no flips | floored+missed now: []
  • dossiq | php unit_size >30 missed->met (23.5%) | floored+missed now: []
  • filinq | php component_independence met->missed (93.8%) FLOORED | floored+missed now: ['php component_independence']
  • hermiq | php module_coupling >10 missed->met (8.4%) FLOORED | floored+missed now: []
  • humaniq | php module_coupling >50 met->missed (1.8%) FLOORED | floored+missed now: ['php module_coupling >50']
  • integriq | php component_independence met->missed (93.7%) FLOORED | floored+missed now: ['php component_independence']
  • keepiq | php module_coupling >10 missed->met (7.4%) FLOORED | floored+missed now: []
  • larpinq | no flips | floored+missed now: []
  • launchpad | no flips | floored+missed now: []
  • learniq | no flips | floored+missed now: []
  • opencatalogi | php component_independence met->missed (94.0%) FLOORED | floored+missed now: ['php component_independence']
  • openregister | no flips | floored+missed now: []
  • pipelinq | no flips | floored+missed now: []
  • planninq | php module_coupling >10 missed->met (7.1%) FLOORED | floored+missed now: []
  • portaliq | php module_coupling >10 missed->met (6.0%) FLOORED | floored+missed now: []
  • shillinq | no flips | floored+missed now: ['vue unit_complexity >5']
  • stackiq | php module_coupling >10 missed->met (9.5%) FLOORED | floored+missed now: []
  • thematiq | php module_coupling >10 missed->met (6.3%) FLOORED | floored+missed now: []
  • versioniq | no flips | floored+missed now: []

Verified

  • test_check_sig_maintainability.py 46 OK, test_check_sig_floor.py 35 OK, test_write_gates_json.py 9 OK (includes a real runner run that writes gates.json).
  • test_gate122_sig_maintainability_scope.sh and test_gate123_sig_floor_scope.sh: all arms passed.
  • Full-scope runner wrote gates.json on humaniq, opencatalogi and buildiq; the other apps are still running.

The page that reads this is the stacked PR for the quality matrix.

AI assistance: written with Claude Code (Opus); reviewed by the submitter before merge.

…tes.json per run

Gate 122 now rates against the 4-star limits of the SIG/TUV NORD CERT
Guidance for Producers v18.1 (2026 model): unit size 48.0/24.2/9.1%,
complexity 20.7/7.5/1.1%, interfacing 15.4/3.4/0.9%, coupling 9.7/5.3/1.8%,
duplication 5.8%, entanglement 0.075, independence 94.5%. Volume is not
rated. Every report row carries the star level its share earns: 5, 3 and 2
stars from the v18.1 limits scaled by the band ratios of Alves, Correia and
Visser 2011 (duplication: Heitlager 2007), 1 star below 2. Independence and
entanglement have 5 stars at half the 4-star exposure and no band under 4.
The five-star profile uses the derived 5-star band instead of half of each
4-star limit.

Gate 123 records the guidance in the floor file and never lowers a floor on
re-recording: a floored check the new limits miss stays in the file and is
named, so the change of status is visible.

The runner writes gates.json next to the findings logs: every declared gate
with its verdict, findings count, scope and matrix category.

Assisted-by: ClaudeCode:claude-opus-5-5
@rubenvdlinde
rubenvdlinde merged commit 238944d into main Nov 6, 2026
57 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant