Framework Architecture

This section contains information about the Metagraph Framework and its relation to the deployed architecture of a metagraph.

In order to understand how the framework functions, it is important to understand the multi-layered architecture of a metagraph. Make sure you're familiar with Metagraph Architecture before continuing.

Code Structure​

A new metagraph project generated from the currency template will have the following module directory structure:

- modules/
- - l0/
- - - Main.scala
- - l1/
- - - Main.scala
- - data_l1/
- - - Main.scala
- - shared_data
- - - Main.scala

Let's break down each directory and its function.

L0​

This directory contains a Main.scala file with a Main object instance that extends CurrencyL0App. CurrencyL0App contains overridable functions that allow for customization of the operation of the metagraph L0 layer including validation, snapshot formation, and management of off chain state. It also contains the rewards overridable function that allows for minting of new tokens on the network.

note

While the class CurrencyL0App has the "Currency" in its name, it defines the L0 layer through which both Currency L1 and Data L1 data flows.

L1​

This directory contains a Main.scala file with a Main object instance that extends CurrencyL1App. CurrencyL1App contains overridable functions relevant to customizing token validation behavior such as transactionValidator.

L1 Data​

This directory also contains a Main.scala file, with a Main object instance that extends CurrencyL1App. In order to have this L1 behave as a DataApplication, the dataApplication method should be overridden with your custom configuration. The metagraph examples repo has implemented examples to reference, for example the NFT example.

Shared Data​

This directory contains an empty Main.scala file but is provided as a suggestion for application directory structure. Several of the lifecycle functions are run on both Data L1 and on L0. For example, serializers/deserializers, validators, and data types will likely shared between layers. Organizing them in a separate directory makes their use in multiple layers clear.

See the Data Application Lifecycle for more information.