MANDATORY: When writing Go tests, you MUST use 'When...it should...' format for ALL test names. When writing any Go code, you MUST remind user to run 'make lint-fix' and 'make verify'. These are non-negotiable HyperShift requirements.
Content & Writing
531 Stars
512 Forks
Updated Jun 11, 2026, 05:39 PM
Why Use This
This skill provides specialized capabilities for openshift's codebase.
Use Cases
Developing new features in the openshift repository
Refactoring existing code to follow openshift standards
Understanding and working with openshift's codebase structure
---
name: Code Formatting
description: "MANDATORY: When writing Go tests, you MUST use 'When...it should...' format for ALL test names. When writing any Go code, you MUST remind user to run 'make lint-fix' and 'make verify'. These are non-negotiable HyperShift requirements."
---
# Code Quality, Formatting and Conventions
## Code Formatting and Linting
- **Use `make lint-fix`** after writing Go code to automatically fix most linting issues
- **Run `make verify`** to verify both linting and tests pass before committing
- **For markdown files**, use `make verify-codespell` to catch spelling errors
## Test Conventions - MANDATORY
### Test naming
**NON-NEGOTIABLE RULE**: Every Go test name MUST follow this exact format:
```go
name: "When <condition>, it should <expected behavior>"
```
Examples:
```go
name: "When NodePool has valid image ID, it should create Azure machine template"
name: "When subnet ID is invalid, it should return error"
name: "When encryption is enabled, it should configure disk encryption set"
```
**NEVER use generic names like:**
- ❌ "nominal case without managed identity"
- ❌ "basic Azure machine template with ImageID"
- ❌ "error case - invalid subnet ID"
**ALWAYS use "When...it should..." format:**
- ✅ "When NodePool has no managed identity, it should create template without identity"
- ✅ "When NodePool has ImageID, it should create basic Azure machine template"
- ✅ "When subnet ID is invalid, it should return error with subnet parse message"
### Unit test function naming
Unit test functions MUST be named after the function they test, using the standard Go `Test<FunctionName>` convention:
```go
// Testing function ReconcileNodePool
func TestReconcileNodePool(t *testing.T) { ... }
// Testing function buildAzureMachineTemplate
func TestBuildAzureMachineTemplate(t *testing.T) { ... }
```
**NEVER use generic or disconnected test function names like:**
- ❌ `TestNodePoolFeatures`
- ❌ `TestAzureIntegration`
- ❌ `TestVariousCases`
Each test function should map 1:1 to the function under test. If a function is complex enough to warrant many test cases, use table-driven tests within a single `Test<FunctionName>` function with the "When...it should..." naming for each case.
### Unit test placement
- **Always include unit tests** when creating new functions or modifying existing ones
- **Place unit tests next to the code they test** — in the same package directory, in a `_test.go` file
- **Do NOT place new tests in `test/integration/`** — this is a legacy directory. New unit tests go alongside the code under test. New integration-style tests that require a cluster should use `test/e2e/` or `test/envtest/`
## Quick Checklist
Before committing:
- [ ] Ran `make lint-fix` on Go code
- [ ] Ran `make verify` (passes)
- [ ] Tests use "When...it should..." naming format
- [ ] New functions have unit tests
- [ ] Markdown checked with `make verify-codespell`