The new KMS-based approach can use a key derivation function that is orders of magnitudes
   cheaper w.r.t. memory and CPU.
- Root credentials can now be changed easily. Before, a two-step process was required to
   change the cluster root credentials since they were used to en/decrypt the IAM data.
   So, both - the old and new credentials - had to be present at the same time during a rotation

## Conditional OpenAPI from settings and env vars

You can easily use the same Pydantic settings to configure your generated OpenAPI and the docs UIs.

For example:

{* ../../docs_src/conditional_openapi/tutorial001.py hl[6,11] *}

 * desirable to use in some unit tests. More importantly, attempting to debug a call which is
 * time-limited would be extremely annoying, so this gives you a time-limiter you can easily swap in
 * for your real time-limiter while you're debugging.
 *
 * @author Kevin Bourrillion
 * @author Jens Nyman
 * @since 1.0
 */
@J2ktIncompatible
@GwtIncompatible
@ElementTypesAreNonnullByDefault

/**
 * Holds a model along with some auxiliary information. This internal utility class assists the model builder during POM
 * processing by providing a means to transport information that cannot be (easily) extracted from the model itself.
 *
 * @deprecated use {@link org.apache.maven.api.services.ModelBuilder} instead
 */
@Deprecated(since = "4.0.0")
class ModelData {
    private final Source source;

# Templates

You can use any template engine you want with **FastAPI**.

A common choice is Jinja2, the same one used by Flask and other tools.

There are utilities to configure it easily that you can use directly in your **FastAPI** application (provided by Starlette).

## Install dependencies

Make sure you create a [virtual environment](../virtual-environments.md){.internal-link target=_blank}, activate it, and install `jinja2`:

<div class="termy">

It was deprecated from Starlette, but if you have code that used it, you can easily **migrate** to <a href="https://github.com/ciscorn/starlette-graphene3" class="external-link" target="_blank">starlette-graphene3</a>, that covers the same use case and has an **almost identical interface**.

/// tip

         * It has an added advantage since you can you easily switch to different hardware accelerators (cpu, gpu, tpu) as per the task.
         * All you need is a good internet connection and you are all set.
      * Try to build TF from sources by changing CPU optimization flags.
   

namespace tensorflow {
namespace checkpoint {

class TensorSliceReader;

// A wrapper around BundleReader (for V2 checkpoints) and
// checkpoint::TensorSliceReader (for V1), that is more easily SWIG wrapped for
// other languages.
//
// The class currently only interacts with single-slice (i.e., non-partitioned)
// variables.
class CheckpointReader {
 public:

# Required-Start: $network $named
# Required-Stop: $network $named
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: This service manages the fess daemon
# Description: Fess is very powerful and easily deployable Enterprise Search Server. You can install and run Fess quickly on any platforms, which have Java runtime environment. Fess is provided under Apache license.
### END INIT INFO

#

	if err != nil {
		log.Fatal(err)
	}
	defer buf.Close()

	if !*flags.SymABIs {
		buf.WriteString(objabi.HeaderString())
		fmt.Fprintf(buf, "!\n")
	}

	// Set macros for GOEXPERIMENTs so we can easily switch
	// runtime assembly code based on them.
	if objabi.LookupPkgSpecial(ctxt.Pkgpath).AllowAsmABI {
		for _, exp := range buildcfg.Experiment.Enabled() {
			flags.D = append(flags.D, "GOEXPERIMENT_"+exp)
		}
	}