Pluggable Serializers for CorDapps heading-link-icon

To be serializable by Corda, Java classes must be compiled with the -parameters switch to enable matching of its properties to constructor parameters. This is important because Corda’s internal AMQP serialization scheme will only construct objects using their constructors. However, when recompilation is not possible, or classes are built in such a way that they cannot be easily modified for simple serialization, CorDapps Corda Distributed Application. A Java (or any JVM targeting language) application built using the Corda build toolchain and CorDapp API to solve some problem that is best solved in a decentralized manner. can provide custom proxy serializers. Corda can use these serializers to move from a type it cannot serialize to an interim representation that it can serialize. The transformation to and from this proxy object is handled by the supplied serializer.

Serializers must:

  • Inherit from net.corda.v5.serialization.SerializationCustomSerializer
  • Provide a proxy class to transform the object to and from
  • Implement the toProxy and fromProxy methods
  • Be included in the CorDapp package Serializers inheriting from SerializationCustomSerializer have to implement two methods and two types.

Consider the following class:

public final class Example {
    private final Int a
    private final Int b

    // Because this is marked private the serialization framework will not
    // consider it when looking to see which constructor should be used
    // when serializing instances of this class.
    private Example(Int a, Int b) {
        this.a = a;
        this.b = b;
    }

    public static Example of (int[] a) { return Example(a[0], a[1]); }

    public int getA() { return a; }
    public int getB() { return b; }
}

Without a custom serializer, we cannot serialize this class as there is no public constructor that facilitates the initialisation of all of its properties.

To be serializable by Corda this would require a custom serializer to be written that can transform the unserializable class into a form we can serialize. Continuing the above example, this could be written as follows:

/**
 * The class lacks a public constructor that takes parameters it can associate
 * with its properties and is thus not serializable by the Corda serialization
 * framework.
 */
class Example {
    private int a;
    private int b;

    public int getA() { return  a; }
    public int getB() { return  b; }

    public Example(List<int> l) {
        this.a = l.get(0);
        this.b = l.get(1);
    }
}

/**
 * This is the class that will Proxy instances of Example within the serializer
 */
public class ExampleProxy {
    /**
     * These properties will be serialized into the byte stream, this is where we choose how to
     * represent instances of the object we're proxying. In this example, which is somewhat
     * contrived, this choice is obvious. In your own classes / 3rd party libraries, however, this
     * may require more thought.
     */
    private int proxiedA;
    private int proxiedB;

    /**
     * The proxy class itself must be serializable by the framework, it must thus have a constructor that
     * can be mapped to the properties of the class via getter methods.
     */
    public int getProxiedA() { return proxiedA; }
    public int getProxiedB() { return proxiedB; }

    public ExampleProxy(int proxiedA, int proxiedB) {
        this.proxiedA = proxiedA;
        this.proxiedB = proxiedB;
    }
}

/**
 * Finally this is the custom serializer that will automatically loaded into the serialization
 * framework when the CorDapp package is loaded at runtime.
 */
public class ExampleSerializer implements SerializationCustomSerializer<Example, ExampleProxy> {

    /**
     *  Given an instance of the Example class, create an instance of the proxying object ExampleProxy.
     *
     *  Essentially convert Example -> ExampleProxy
     */
    public ExampleProxy toProxy(Example obj) {
        return new ExampleProxy(obj.getA(), obj.getB());
    }

    /**
     * Conversely, given an instance of the proxy object, revert that back to an instance of the
     * type being proxied.
     *
     *  Essentially convert ExampleProxy -> Example
     */
    public Example fromProxy(ExampleProxy proxy) {
        List<int> l = new ArrayList<int>(2);
        l.add(proxy.getProxiedA());
        l.add(proxy.getProxiedB());
        return new Example(l);
    }
}
class ExampleSerializer : SerializationCustomSerializer<Example, ExampleSerializer.Proxy> {
    /**
     * This is the actual proxy class that is used as an intermediate representation
     * of the Example class
     */
    data class Proxy(val a: Int, val b: Int)

    /**
     * This method should be able to take an instance of the type being proxied and
     * transpose it into that form, instantiating an instance of the Proxy object (it
     * is this class instance that will be serialized into the byte stream.
     */
    override fun toProxy(obj: Example) = Proxy(obj.a, obj.b)

    /**
     * This method is used during deserialization. The bytes will have been read
     * from the serialized blob and an instance of the Proxy class returned, we must
     * now be able to transform that back into an instance of our original class.
     *
     * In our example this requires us to evoke the static "of" method on the
     * Example class, transforming the serialized properties of the Proxy instance
     * into a form expected by the construction method of Example.
     */
    override fun fromProxy(proxy: Proxy) : Example {
        val constructorArg = IntArray(2);
        constructorArg[0] = proxy.a
        constructorArg[1] = proxy.b
        return Example.of(constructorArg)
    }
}

In the above examples

  • ExampleSerializer is the actual serializer that will be loaded by the framework to serialize instances of the Example type.
  • ExampleSerializer.Proxy, in the Kotlin example, and ExampleProxy, in the Java example, is the intermediate representation used by the framework to represent instances of Example within the wire format.

The proxy object should be thought of as an intermediate representation that the serialization framework can reason about. One is being written for a class because, for some reason, that class cannot be introspected successfully but that framework can only contain elements that the framework can reason about.

The proxy class itself is distinct from the proxy serializer. The serializer must refer to the unserializable type in the toProxy and fromProxy methods.

For example, the first thought a developer may have when implementing a proxy class is to simply wrap an instance of the object being proxied. This is shown below:

class ExampleSerializer : SerializationCustomSerializer<Example, ExampleSerializer.Proxy> {
    /**
     * In this example, we are trying to wrap the Example type to make it serializable
     */
    data class Proxy(val e: Example)

    override fun toProxy(obj: Example) = Proxy(obj)

    override fun fromProxy(proxy: Proxy) : Example {
        return proxy.e
    }
}

However, this will not work because what we’ve created is a recursive loop whereby synthesising a serializer for the Example type requires synthesising one for ExampleSerializer.Proxy. However, that requires one for Example and so on until we get a StackOverflowException.

The solution, as shown initially, is to create the intermediate form (the Proxy object) purely in terms which the serialization framework can reason about.

Writing a custom serializer for a class has the effect of adding that class to the allow list, meaning such classes do not need the @CordaSerializable annotation.

Was this page helpful?

Thanks for your feedback!

Chat with us

Chat with us on our #docs channel on slack. You can also join a lot of other slack channels there and have access to 1-on-1 communication with members of the R3 team and the online community.

Propose documentation improvements directly

Help us to improve the docs by contributing directly. It's simple - just fork this repository and raise a PR of your own - R3's Technical Writers will review it and apply the relevant suggestions.

We're sorry this page wasn't helpful. Let us know how we can make it better!

Chat with us

Chat with us on our #docs channel on slack. You can also join a lot of other slack channels there and have access to 1-on-1 communication with members of the R3 team and the online community.

Create an issue

Create a new GitHub issue in this repository - submit technical feedback, draw attention to a potential documentation bug, or share ideas for improvement and general feedback.

Propose documentation improvements directly

Help us to improve the docs by contributing directly. It's simple - just fork this repository and raise a PR of your own - R3's Technical Writers will review it and apply the relevant suggestions.