Speed up CI/CD runs, scope the developer CLI, and harden workflow hygiene

[tjementum]

Summary & Motivation

A coordinated refresh of the CI/CD pipelines and developer CLI: faster runs, scoped commands, better observability, and consistent workflow conventions.

Test infrastructure

BackOfficeEndpointBaseTest and EndpointBaseTest<TContext> built a fresh WebApplicationFactory<Program> per test instance. With Meziantou.Xunit.ParallelTestFramework, every Account-API and BackOffice test paid a full ASP.NET Core cold-start. Both bases now share one host per test class via IClassFixture<...>, routing per-test state (SqliteConnection, telemetry collector, MockStripeState, IEmailClient) through AsyncLocal. New types: BackOfficeWebApplicationFactory + BackOfficeTestContext (26 derived classes) and AccountWebApplicationFactory + AccountTestContext (60 derived classes). MockStripeState becomes transient so per-request resolution reads the per-test instance; TestServer.PreserveExecutionContext = true keeps the AsyncLocal flowing into request handling.

Single Code Style workflow

A new code-style.yml replaces the per-SCS code-style work in account.yml, main.yml, app-gateway.yml. Four jobs:

• detect-scope — classifies the diff via inline git diff, outputs the backend scope and format mode for the downstream jobs.
• code-linting — backend inspectcode + frontend build + frontend lint (oxlint).
• code-formatting — backend cleanupcode + frontend format check (oxfmt).
• sonarcloud — runs once per push (previously three times, once per SCS), sequentially per slnf to keep test assemblies in separate processes and avoid the PortAllocation.Load() race.

Backend lint and format are scoped to a single SCS via --self-contained-system <name> when the diff touches only that SCS; cross-cutting changes fall back to the full solution. Code coverage is out of scope here — the dotCover path repeatedly exceeded 20 minutes against the new IClassFixture infrastructure and is tracked separately.

AppGateway tests now run in CI

AppGateway.Tests existed in the solution but was never executed — the slnf-scoped runs missed it, and app-gateway.yml had no test step. The workflow now runs dotnet test AppGateway.Tests/AppGateway.Tests.csproj --no-build with the same user-secret bootstrap the other workflows use.

Developer CLI: --gateway/-g scope flag

build, test, format, and lint accept --gateway/-g to scope to AppGateway only (mutually exclusive with -s). pp test -g targets AppGateway.Tests/AppGateway.Tests.csproj; lint/format scope JetBrains tools to AppGateway/** + AppGateway.Tests/**. Shared logic in AppGatewayHelper.

Faster format and lint

• format defaults to changed-only (.cs files diffed vs origin/main); --all-files for full sweep. CI auto-flips to --all-files when application/dotnet-tools.json changes, so JetBrains tool upgrades catch latent drift.
• lint --changed-only is opt-in; CI lints full because inspectcode has cross-file rules.
• Dropped the temp-.slnf workaround for cleanupcode — JetBrains 2026.1 accepts .slnx directly.

Workflow hygiene

• Concurrency on every top-level workflow with cancel-in-progress: true (and false for reusable deploy workflows so mid-deploy cancellations cannot leave a half-applied state).
• Inspection results on failure: code-style.yml and developer-cli.yml dump result.json to the run console and $GITHUB_STEP_SUMMARY. Replaces the brittle grep | tee exit-code contract.
• Removed leftover SonarScanner steps and Java JDK setup from account.yml and main.yml (centralized in code-style.yml).
• Removed stale _preview-migrations.yml trigger-path references (file does not exist).
• YAML hygiene pass across all workflows: blank lines, canonical argument order, trailing whitespace.

Database Plan reuses Build and Test artifacts

build-and-test uploads application/**/bin and application/**/obj as a <scs>-build artifact (when staging is enabled). _migrate-database.yml downloads it and restores NuGet, skipping the redundant checkout + setup + build chain.

E2E test fix piggy-backed on this branch

Two e2e tests inherited from the recently-merged feature-flags work failed on a default-seeded database (feature-flag-flows.spec.ts:262, user-management-flows.spec.ts:473). Root cause: account-overview and compact-view are registered with isKillSwitchEnabled: true, so the startup reconciler creates them inactive (EnabledAt = null); the configurable-flags handlers filter inactive rows out, the API returns empty, and FeaturesSection / PreferencesFeatureFlagsSection render null. The tests now activate the kill-switch flags via the back-office activate endpoint during setup — test-only fix, no production code, no seeder change.

CI benchmark

The biggest CI win comes from the Database Staging migration-skip gate on account.yml / main.yml — PRs that do not change EF migration files skip the Database Staging job entirely.

Workflow	BEFORE	AFTER, no migrations	AFTER, with migrations	Δ (no migrations)	Δ (with migrations)
Account	9m00s	2m49s	6m17s	-69%	-30%
Main	6m41s	1m55s	5m40s	-71%	-15%
AppGateway	7m26s	1m26s	1m26s	-81%	-81%
Code Style	-	4m48s	4m48s	new	new
Critical path per push	9m00s	4m48s	6m17s	-47%	-30%
Total CI minutes per push	23m07s	15m19s	22m34s	-34%	-2%

AppGateway's delta is large because code-linting and code-formatting moved out of app-gateway.yml into the consolidated code-style.yml. On the no-migrations path the critical path now sits on Code Style (parallel lint/format/sonar jobs each paying setup); sharing the backend build artifact across them is an obvious follow-up. On the with-migrations path the critical path stays on account.yml.

Downstream projects with a substantial Main SCS will see a bigger lift than this suggests — the slowest CI step used to be backend cleanupcode running on every file; it now inspects only changed .cs files. Main and Account now sit close in wall-clock despite Account having materially more code, because the npm + dotnet restore + build setup now dominates.

Test-step benchmark (isolated)

Test-host refactor measured in isolation (Code Style and Database Staging disabled; 5 samples per side):

	Run 1	Run 2	Run 3	Run 4	Run 5	Mean	Stddev
AFTER (BackOffice + Account shared host)	2m18s	2m26s	2m21s	2m21s	2m21s	2m21s	2.9s
BEFORE (per-test host, baseline)	3m46s	3m51s	3m38s	3m41s	3m54s	3m46s	6.7s

Δ -37% on the test step alone, stddev more than halves — the runs are markedly more stable too.

Checklist

• I have added tests, or done manual regression tests
• I have updated the documentation, if necessary

[linear]

PP-1209

PP-1262

PP-809

PP-1261

PP-951

PP-1217

[sonarqubecloud]

Quality Gate passed

Issues
13 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud