# ADR-0002 - Avoid using Java serialization

## Date

2012-12-01

## Context

In Gradle we often need to serialize in-memory objects for caching, or to transmit them across process barriers, etc.
Java serialization is one way to implement this, however, despite its simplicity of implementation, it has several drawbacks:

- **Performance:**
Java's built-in serialization mechanism is often slower compared to other serialization solutions.

     * @param columnNumber The one-based index of the column containing the error or {@code -1} if unknown.
     * @param exception The exception that caused this problem, may be {@code null}.
     */
    // mkleint: does this need to be public?
    public DefaultModelProblem(
            String message,
            Severity severity,
            Version version,
            Model source,
            int lineNumber,
            int columnNumber,

`Cluster` is pushed, so this optimization is intentionally turned off in this specific case.

Finally, we determine which subset of the type we need to generate. XDS has two modes - "State of the World (SotW)" and "Delta". In SotW, we generally need to generate all resources of the type, even if only one changed. Note that we actually need to *generate* all of them, typically, as we do not store previously generated resources (mostly because they are generated per-client). This also means...

//
// Gradient is the type returned by gradient functions. In Python TF it's either
// Tensor or IndexedSlices or None, which here we map to nullptr. Gradients need
// to allow their size to be computed and they need to be passable to a backward
// function and deleted (as the backprop code creates lots of gradients the user
// is not interested in).
//

]
```

These are **JavaScript** objects, not strings, so you can't pass them from Python code directly.

# The `automations` section lists automations to run on PRs in this repository.
# Each automation has an `if` key that defines conditions to run the automation,
# as well as a `run` key with the actions to do. All the conditions need to be true
# for the actions to run (there is an implicit AND relation between
# the conditions on each line).  Each condition is specified as a Jinja expression.  You

!!! info
    Keep in mind that dataclasses can't do everything Pydantic models can do.

    So, you might still need to use Pydantic models.

    But if you have a bunch of dataclasses laying around, this is a nice trick to use them to power a web API using FastAPI. 🤓

## Dataclasses in `response_model`

For distributed MinIO deployments, specify the *same* `MINIO_KMS_SECRET_KEY` for each MinIO server process.

At any point in time you can switch from `MINIO_KMS_SECRET_KEY` to a full KMS
deployment. You just need to import the generated key into KES - for example via
the KES CLI once you have successfully setup KES:

```sh
kes key create my-minio-key OSMM+vkKUTCvQs9YL/CVMIMt43HFhkUpqJxTmGl6rYw=
```

And then you can set headers in that *temporal* response object.

```Python hl_lines="1  7-8"
{!../../../docs_src/response_headers/tutorial002.py!}
```

And then you can return any object you need, as you normally would (a `dict`, a database model, etc).

And if you declared a `response_model`, it will still be used to filter and convert the object you returned.

      render: java
    validations:
      required: false

  - type: markdown
    attributes:
      value: >
        #### Ubiquity: provide concrete use cases


        Did you *actually* encounter the need for this feature in a real-world scenario, or is it
        just a feature that seems like a sensible addition to Guava?


        Before new features get added to Guava, we really want to be sure that it's for a use case