Quality and Metrics

Quality and Metrics

SharpAccess retains exact-revision engineering evidence for the supported production scope: SharpAccess.Core, SharpAccess.Sqlite, and SharpAccess.Postgres.

The block below is generated from the same artifacts/quality-report/metrics.json source as the root README. Do not edit the metric values manually.

Source: exact-revision artifacts/quality-report/metrics.json · Schema: 2 · Enforcement: EvidenceOnly

Metric	p95	Worst observed	Aggregate / release interpretation
Line coverage	100.00%	0.00% minimum	91.39% repository aggregate
Branch coverage	100.00%	0.00% minimum	81.99% repository aggregate
CRAP score	12.14	25.48 maximum	Executable methods only
Cyclomatic complexity	8	24 maximum	Roslyn source metrics
Maintainability index	100	41 minimum	Higher is better
Class coupling	15	49 maximum	Distinct referenced types
Afferent coupling (Ca)	11.6	13 maximum	Project + namespace units
Efferent coupling (Ce)	8	9 maximum	Project + namespace units
Instability	1.000	1.000 maximum	Ce / (Ca + Ce); informational
Critical mutation invariants	N/A	0 survived; 0 infrastructure failures required	Binary per selected invariant; release tier must pass

Scope and percentile notes

• Coverage percentiles are calculated across matched executable production members in SharpAccess.Core, SharpAccess.Sqlite, and SharpAccess.Postgres.
• The repository aggregate remains the release coverage score; the minimum exposes the worst observed member because coverage is higher-is-better.
• CRAP, cyclomatic complexity, maintainability index, and class-coupling statistics use the report's exact member dataset.
• Ca, Ce, and instability are calculated across project and namespace dependency units.
• Mutation invariants are binary and therefore do not have a meaningful p95.
• Consult artifacts/quality-report/index.html and metrics.json for complete project, namespace, type, member, dependency, and hotspot detail.

Blocking quality gates

• Core coverage: at least 85% line and 75% branch.
• SQLite coverage: at least 80% line and 65% branch.
• PostgreSQL coverage: at least 80% line and 65% branch while Supported.
• Changed handwritten production code: at least 90% line and 75% branch.
• Supported combined coverage retains an additional checked-in non-regression floor.
• New complexity/CRAP hotspots are rejected.
• Existing reviewed hotspots may not gain cyclomatic complexity or exceed the configured CRAP tolerance.
• Every selected critical mutation must be killed.
• Missing required evidence fails closed.

Report outputs

Complete local verification produces:

artifacts/quality-report/index.html
artifacts/quality-report/metrics.json
artifacts/quality-report/manifest.json

The report includes project, namespace, type, member, dependency, and hotspot views.

Interpretation

Coverage and maintainability are higher-is-better, so the worst observed member is the minimum. CRAP, cyclomatic complexity, coupling, and instability are adverse metrics, so the worst observed value is the maximum. Mutation invariants are binary and do not have a meaningful percentile.

Commands

./scripts/verify-local.ps1 -RepositoryRoot $PWD -Version '0.9.0-rc.1'
./scripts/quality-report.ps1 -RepositoryRoot $PWD
./scripts/complexity-report.ps1 -RepositoryRoot $PWD
./scripts/mutation-test.ps1 -RepositoryRoot $PWD

Use the exact repository-supported parameters for the selected workflow.

References

• Quality gates
• Quality report
• Coverage and test gates
• Verification and complexity