Skip to main content

Where LionWeb Nodes and Languages Live?

In-Memory Only​

Nodes can exist purely in memory. This is typical for short-lived operations:

  • a batch process creates some nodes,
  • performs analysis or transformation,
  • produces an artifact (e.g., a report),
  • discards the nodes when the operation ends.

In this scenario, nodes are regular in-memory objects created through any LionWeb library, without persistence.

Persistent Repository​

The most common setup is to store nodes in a persistent repository.
A repository:

  • keeps nodes and partitions over time,
  • offers services for accessing, querying, and modifying data,
  • acts as the authoritative storage for tools that rely on shared models.

This is the scenario most LionWeb-based systems eventually adopt.

Files on Disk (JSON or Protobuf)​

Nodes can also be persisted as files:

  • JSON is supported by all LionWeb libraries.
  • Protobuf is supported by most libraries.

You can:

  • store each partition in its own file, or
  • use the archive format (a ZIP container holding multiple encoded partitions).

The archive format is currently available in the Java and Python implementations.
It groups multiple partitions and languages into a single compressed file containing protobuf-encoded data.

Storing Languages​

Languages are nodes too. All storage mechanisms described above apply to languages the same way they apply to regular nodes.

Some considerations:

  • A specialized editor might embed a language definition directly.
  • A language might be shipped as a file alongside the tool.
  • A language may also be constructed programmatically when the editor starts.

These choices depend on how dynamic the language needs to be and how the tool is distributed.