However, there are some things to consider depending on whether you need module information to be available, and module boundaries to be enforced, during test execution.
In this context, the terms _whitebox testing_ (module boundaries are deactivated or relaxed) and _blackbox testing_ (module boundaries are in place) are often used.
Whitebox testing is used/needed for unit testing and blackbox testing fits functional or integration test requirements.

func applyRewrite(f *Func, rb blockRewriter, rv valueRewriter, deadcode deadValueChoice) {
	// repeat rewrites until we find no more rewrites
	pendingLines := f.cachedLineStarts // Holds statement boundaries that need to be moved to a new value/block
	pendingLines.clear()
	debug := f.pass.debug
	if debug > 1 {
		fmt.Printf("%s: rewriting for %s\n", f.pass.name, f.Name)
	}

// On a related note, small objects are always small enough that their bitmap
// fits in goarch.PtrSize*8 bits, so writing out bitmap data takes two bitmap
// writes at most (because object boundaries don't generally lie on
// s.heapBits()[i] boundaries).
//
// For larger objects, if t is the type for the object starting at "start",
// within some span whose mspan is s, then the bitmap at t.GCData is "tiled"

	// of the steps we need with a single traceback printer implementation.
	//
	// We could be more lax about exactly how many frames we print, for example
	// always stopping and resuming on physical frame boundaries, or at least
	// cgo expansion boundaries. It's not clear that's much simpler.
	flags |= unwindPrintErrors
	var u unwinder
	tracebackWithRuntime := func(showRuntime bool) int {
		const maxInt int = 0x7fffffff

  Status GetFunctionNameAttr(Node const* node, string* attr) const;

  // Copies edges local to a subgraph. Adds _Arg and _Retval nodes to
  // subgraphs for data edges that cross subgraph boundaries.
  Status CopySubgraphEdges(
      const absl::flat_hash_map<const Node*, Node*>& node_images,
      std::vector<std::pair<const Node*, Node*>>* src_arg_pairs);

	sc.setNonEmpty()
}

type piController struct {
	kp float64 // Proportional constant.
	ti float64 // Integral time constant.
	tt float64 // Reset time.

	min, max float64 // Output boundaries.

	// PI controller state.

	errIntegral float64 // Integral of the error from t=0 to now.

	// Error flags.
	errOverflow   bool // Set if errIntegral ever overflowed.

	errs := make(chan error, workersNumber+1)
	workersExited := make(chan struct{})

	wg.Add(workersNumber)
	for i := 0; i < workersNumber; i++ {
		go func() {
			// panics don't cross goroutine boundaries
			defer utilruntime.HandleCrash(func(panicReason interface{}) {
				errs <- fmt.Errorf("DeleteCollection goroutine panicked: %v", panicReason)
			})
			defer wg.Done()

			for item := range toProcess {

//
// The calendar runs on an exact 400 year cycle: a 400-year calendar
// printed for 1970-2369 will apply as well to 2370-2769. Even the days
// of the week match up. It simplifies the computations to choose the
// cycle boundaries so that the exceptional years are always delayed as
// long as possible. That means choosing a year equal to 1 mod 400, so
// that the first leap year is the 4th year, the first missed leap year

[[sec:interoperability]]
== Interoperability

When mixing languages in your build logic, you may have to cross language boundaries.
An extreme example would be a build that uses tasks and plugins that are implemented in Java, Groovy and Kotlin, while also using both Kotlin DSL and Groovy DSL build scripts.

Quoting the Kotlin reference documentation:

	pph.Paddr = uint64(*FlagTextAddr) - uint64(HEADR) + pph.Off
	pph.Align = uint64(*FlagRound)

	/*
	 * PHDR must be in a loaded segment. Adjust the text
	 * segment boundaries downwards to include it.
	 */
	{
		o := int64(Segtext.Vaddr - pph.Vaddr)
		Segtext.Vaddr -= uint64(o)
		Segtext.Length += uint64(o)
		o = int64(Segtext.Fileoff - pph.Off)
		Segtext.Fileoff -= uint64(o)