return errors.NewInternalError(fmt.Errorf("Expected Ingress resource, got: %v", attr.GetKind()))
	}

	// IngressClassName field has been set, no need to set a default value.
	if ingress.Spec.IngressClassName != nil {
		return nil
	}

	// Ingress class annotation has been set, no need to set a default value.
	if _, ok := ingress.Annotations[networkingv1beta1.AnnotationIngressClass]; ok {
		return nil
	}

need it so that empty directory does not get deleted...

## Configuration protocol

Ztunnel, of course, needs to be dynamically configured in order to make decisions on how it should handle traffic.
For this purpose, we chose to use xDS transport protocol due to our expertise and existing infrastructure, and because the protocol is well suited to our needs.

     * These classes are deleted before a compilation and are not passed to Java compiler (their sources are passed to a compiler).
     * We only need them in {@link CompilationClassBackupService} so we know what files don't need a backup.
     */
    Set<String> getClassesToCompile();

    List<File> getModulePath();

    void setModulePath(List<File> modulePath);

#include "tensorflow/compiler/mlir/lite/experimental/tac/common/targets.h"

namespace mlir {
namespace TFL {
namespace tac {

// TODO(renjieliu): We need to come up with a better strategy to do cost
// estimatation. Maybe build a big lookup table for all the ops.

// TODO(renjieliu): We need to consider what's the default value if we cannot
// analyze the cost.

// ================== Interface  ========================

                        // holds without causing a deadlock. This in turn implies that this thread does not need to release the project locks it holds while
                        // blocking waiting for an operation to complete and does not need to deal with another thread stealing its project lock(s) while blocking.
                        //

So we have a dedicated thread for every socket that just reads frames and dispatches them.

This can be very useful for setting up **resources** that you need to use for the whole app, and that are **shared** among requests, and/or that you need to **clean up** afterwards. For example, a database connection pool, or loading a shared machine learning model.

## Use Case

make docker

# Push images to docker registry
make push
```

On MacOS, you need to set the target operating system before building the images

```bash
GOOS=linux make docker push
```

### Option 2: Use pre-built Istio images

 *
 * <p>This class is used only for micro-optimization. Standard {@code Future} utilities benefit from
 * this optimization, so there is no need to specialize methods to return or accept this type
 * instead of {@code ListenableFuture}.
 *
 * <p>This class is GWT-compatible.
 *