jattac.libs.web.zest-button 1.2.9 → 1.4.0

package/docs/guidelines/AI_TESTING.md

@@ -0,0 +1,275 @@
+# Testing Policy
+
+This repository uses strict Test Driven Development.
+
+No exceptions.
+
+---
+
+# Fundamental Rule
+
+Production code MUST NOT be written before a failing test exists.
+
+---
+
+# Required Cycle
+
+Write failing test
+
+↓
+
+Observe failure
+
+↓
+
+Write minimal production code
+
+↓
+
+Observe success
+
+↓
+
+Refactor
+
+---
+
+# Bug Fixes
+
+Every bug MUST begin with a regression test.
+
+Never fix a bug first.
+
+Incorrect:
+
+Fix bug
+
+↓
+
+Write test
+
+Correct:
+
+Write failing regression test
+
+↓
+
+Observe failure
+
+↓
+
+Implement fix
+
+↓
+
+Observe pass
+
+---
+
+# Existing Behaviour
+
+Before modifying existing code:
+
+Identify all affected behaviour.
+
+If behaviour is not covered by tests:
+
+Write those tests first.
+
+The objective is to freeze current business behaviour.
+
+Only then modify production code.
+
+---
+
+# Missing Tests
+
+Missing tests are technical debt.
+
+If encountered during implementation they SHOULD be added before production changes.
+
+---
+
+# Green Tests
+
+Code SHALL NEVER be committed while tests are failing unless explicitly instructed.
+
+---
+
+# Refactoring
+
+Refactoring SHALL NOT alter externally observable behaviour.
+
+Tests SHALL remain green before, during and after refactoring.
+
+---
+
+# Coverage
+
+Coverage percentage is NOT success.
+
+Correct behavioural coverage is success.
+
+Prefer meaningful behavioural tests over line coverage.
+
+---
+
+# Coverage Regression
+
+Coverage MUST NOT decrease from the previous build.
+
+If current coverage is lower than the baseline, the build FAILS.
+
+This is not about hitting a target number.
+
+This is about preventing regressions.
+
+The baseline is tracked in test-reports/coverage-baseline.json.
+
+See AI_TEST_CONFIGURATION.md for baseline management details.
+
+---
+
+# Purpose of Tests
+
+Tests are written to verify functionality and catch bugs.
+
+Tests are NOT written to hit coverage goals.
+
+High coverage with poor tests is worse than lower coverage with meaningful tests.
+
+Every test MUST verify a specific behaviour.
+
+Every test MUST assert specific expected values.
+
+Coverage is a side effect of good testing, not the goal.
+
+---
+
+# Test Reporting Configuration
+
+Every project MUST have test reporting configured.
+
+The LLM MUST verify configuration at the start of every session.
+
+If reporting is not configured, the FIRST task is to configure it.
+
+See AI_TEST_CONFIGURATION.md for platform-specific instructions.
+
+---
+
+# Test Output Verification
+
+When tests are run, you MUST capture and report the actual output.
+
+State:
+
+• Command run
+
+• Tests passed
+
+• Tests failed
+
+• Tests skipped
+
+Do NOT summarise.
+
+Do NOT paraphrase.
+
+Show the numbers.
+
+Example:
+
+```
+$ dotnet test
+
+Passed:  47
+Failed:  0
+Skipped: 2
+
+Total:   49
+```
+
+Do NOT claim tests pass without showing output.
+
+---
+
+# Failure Handling
+
+If ANY test fails, the task is NOT complete.
+
+Fix the failure before proceeding.
+
+Do NOT skip tests.
+
+Do NOT disable tests.
+
+Do NOT mark tests as ignored.
+
+Do NOT modify tests to make them pass.
+
+Do NOT change test expectations to match broken behaviour.
+
+---
+
+# Test Isolation
+
+New tests MUST NOT depend on test execution order.
+
+New tests MUST NOT depend on external state (databases, file systems, network) unless testing integration behaviour explicitly.
+
+New tests MUST clean up after themselves.
+
+New tests MUST NOT modify shared state that could affect other tests.
+
+---
+
+# Assertion Specificity
+
+Tests MUST assert specific expected values, not just "does not throw."
+
+Tests MUST verify behaviour, not implementation details.
+
+Tests SHOULD use Arrange-Act-Assert structure.
+
+Tests SHOULD have descriptive names that explain the scenario.
+
+---
+
+# Test Categories
+
+The following test categories exist and MUST be executed in order:
+
+1. Unit tests
+
+2. Service tests
+
+3. API integration tests
+
+4. API contract tests
+
+5. UI end-to-end tests
+
+All categories MUST pass before a task is considered complete.
+
+If a test category does not exist in the project, skip it but document its absence.
+
+---
+
+# Mock and Stub Rules
+
+Mocks and stubs MUST NOT be used to hide failures.
+
+Mocks and stubs MUST represent real behaviour accurately.
+
+If a mock is causing confusion, write an integration test instead.
+
+---
+
+# Regression Tests
+
+Every bug fix MUST include a regression test.
+
+Regression tests MUST be named to describe the bug they prevent.
+
+Regression tests MUST remain in the test suite permanently.