Go Linters

Build and test linters

• Test only your linter package:
  • go test ./pkg/linters/<linter-name>/...
• Build the custom linter runner:
  • go build ./cmd/linters
• Run all custom linters across the repo:
  • make golint-custom

make golint-custom builds cmd/linters and runs it against ./cmd/... and ./pkg/....

Coverage-aware perf gating

For linters that flag micro-optimizations (allocation/perf rules), only apply them on lines that tests actually exercise — “hot paths” — rather than on dead or rarely-executed code where the optimization brings no measurable benefit. Use the shared pkg/linters/internal/coverage package:

1. In your analyzer file, register a -hot-threshold flag in init() (not as a var initializer, to avoid an Analyzer/run/flag initialization cycle):

   goCopy

   var hotThreshold *int
   
   func init() {
       hotThreshold = coverage.RegisterHotThresholdFlag(Analyzer)
   }

2. Immediately before reporting a diagnostic, gate it with coverage.ShouldApply:

   goCopy

   if !coverage.ShouldApply(pass, node.Pos(), *hotThreshold) {
       return
   }

coverage.ShouldApply is permissive by default: when no coverage profile is loaded via the GH_AW_LINT_COVERAGE_PROFILE environment variable, or when hot-threshold is 0, it always returns true, preserving pre-coverage-aware behavior. Only wire this into linters whose fix has a genuine performance rationale (extra allocations, O(n²) behavior, etc.) — purely readability/style linters should not be coverage-gated.

Generating the coverage profile

This profile is read once per linter-runner process. To lint only a specific subtree, scope the go test and golint-custom commands to the same package path.