+++ title = "Codebase overview" template = "book.html" page_template = "book.html" [extra] order = 2 # Chapter number js = ["/js/youtube-embed.js"] css = ["/component/youtube-embed.css"] +++ The best introduction for getting up-to-speed with Graphite contribution comes from watching this webcast recording. Before asking questions in Discord, please watch the full video because it gives a comprehensive overview of most things you will need to know.
Workshop: Intro to Coding for Graphite
## Codebase structure Graphite is built from several main software components. New developers may choose to specialize in one or more area without having to attain a working knowledge of the full codebase. ### Frontend *Location: `/frontend/src`* The frontend is the interface for Graphite which users see and interact with. It is built using web technologies with TypeScript and Svelte (HTML and SCSS). The frontend's philosophy is to be as lightweight and minimal as possible. It acts as the entry point for user input and then quickly hands off its work to the WebAssembly editor backend via its Wasm wrapper API. That API is written in Rust but has TypeScript bindings generated by the [wasm-bindgen](https://github.com/rustwasm/wasm-bindgen) tooling that is part of the Vite-based build chain. The frontend is built of many components that recursively form the window, panels, and widgets that make up the user interface. ### Editor *Location: `/editor`* The editor is the core of the Graphite application, and it's where all the business logic occurs for the tooling and user interaction. It is written in Rust and compiled to WebAssembly. At its heart is the message system described below. It is responsible for communicating with Graphene as well as handling the actual logic, state, tooling, and responsibilities of the interactive application. ### Graphene *Location: `/node-graph`* [Graphene](../graphene/) is the node graph engine which manages and renders the documents. It is itself a programming language, where Graphene programs are compiled while being edited live by the user, and where executing the program renders the document. ## Frontend/backend communication Frontend-to-backend communication is achieved through a thin Rust translation layer in `/frontend/wasm/src/editor_api.rs` which wraps the editor backend's Rust-based message system API and provides the TypeScript-compatible API of callable functions. These wrapper functions are compiled by [wasm-bindgen](https://github.com/rustwasm/wasm-bindgen) into autogenerated TS functions that serve as an entry point from TS into the Wasm binary. Backend-to-frontend communication happens by sending a queue of messages to the frontend message dispatcher. After the TS has called any wrapper API function to get into backend code execution, the editor's business logic runs and queues up `FrontendMessage`s (defined in `/editor/src/messages/frontend/frontend_message.rs`) which get mapped from Rust to TS-friendly data types in `/frontend/src/messages.ts`. Various TS code subscribes to these messages by calling `subscribeJsMessage(MessageName, (messageData) => { /* callback code */ });`. ## The message system The Graphite editor backend is organized into a hierarchy of subsystems, called *message handlers*, which talk to one another through message passing. Messages are pushed to the front or back of a queue and each one is processed sequentially by the backend's dispatcher. The dispatcher lives at the root of the application hierarchy and it owns its message handlers. Thus, Rust's restrictions on mutable borrowing are satisfied because only the dispatcher mutably borrows its message handlers, one at a time, while each message is processed. ### Messages Messages are enum variants that are dispatched to perform some intended activity within their respective message handlers. Here are two `DocumentMessage` definitions: ```rs pub enum DocumentMessage { ... // A message that carries one named data field DeleteLayer { id: NodeId, } // A message that carries no data DeleteSelectedLayers, ... } ``` As shown above, additional data fields can be included with each message. But as a special case denoted by the `#[child]` attribute, that data can also be a sub-message enum, which enables hierarchical nesting of message handler subsystems.
To view the hierarchical subsystem file structure: click here ``` messages ├── broadcast │   ├── broadcast_message.rs │   └── broadcast_message_handler.rs ├── debug │   ├── debug_message.rs │   └── debug_message_handler.rs ├── dialog │   ├── dialog_message.rs │   ├── dialog_message_handler.rs │   ├── export_dialog │   │   ├── export_dialog_message.rs │   │   └── export_dialog_message_handler.rs │   ├── new_document_dialog │   │   ├── new_document_dialog_message.rs │   │   └── new_document_dialog_message_handler.rs │   └── preferences_dialog │   ├── preferences_dialog_message.rs │   └── preferences_dialog_message_handler.rs ├── frontend │   └── frontend_message.rs ├── globals │   ├── globals_message.rs │   └── globals_message_handler.rs ├── input_mapper │   ├── input_mapper_message.rs │   ├── input_mapper_message_handler.rs │   └── key_mapping │   ├── key_mapping_message.rs │   └── key_mapping_message_handler.rs ├── input_preprocessor │   ├── input_preprocessor_message.rs │   └── input_preprocessor_message_handler.rs ├── layout │   ├── layout_message.rs │   └── layout_message_handler.rs ├── portfolio │   ├── document │   │   ├── document_message.rs │   │   ├── document_message_handler.rs │   │   ├── graph_operation │   │   │   ├── graph_operation_message.rs │   │   │   └── graph_operation_message_handler.rs │   │   ├── navigation │   │   │   ├── navigation_message.rs │   │   │   └── navigation_message_handler.rs │   │   ├── node_graph │   │   │   ├── node_graph_message.rs │   │   │   └── node_graph_message_handler.rs │   │   ├── overlays │   │   │   ├── overlays_message.rs │   │   │   └── overlays_message_handler.rs │   │   └── properties_panel │   │   ├── properties_panel_message.rs │   │   └── properties_panel_message_handler.rs │   ├── menu_bar │   │   ├── menu_bar_message.rs │   │   └── menu_bar_message_handler.rs │   ├── portfolio_message.rs │   └── portfolio_message_handler.rs ├── preferences │   ├── preferences_message.rs │   └── preferences_message_handler.rs ├── tool │   ├── tool_message.rs │   ├── tool_message_handler.rs │   ├── tool_messages │   │   ├── artboard_tool.rs │   │   ├── brush_tool.rs │   │   ├── ellipse_tool.rs │   │   ├── eyedropper_tool.rs │   │   ├── fill_tool.rs │   │   ├── freehand_tool.rs │   │   ├── gradient_tool.rs │   │   ├── line_tool.rs │   │   ├── navigate_tool.rs │   │   ├── path_tool.rs │   │   ├── pen_tool.rs │   │   ├── polygon_tool.rs │   │   ├── rectangle_tool.rs │   │   ├── select_tool.rs │   │   ├── spline_tool.rs │   │   └── text_tool.rs │   └── transform_layer │   ├── transform_layer_message.rs │   └── transform_layer_message_handler.rs └── workspace ├── workspace_message.rs └── workspace_message_handler.rs ```
By convention, regular data must be written as struct-style named fields (shown above), while a sub-message enum must be written as a tuple/newtype-style field (shown below). The `DocumentMessage` enum of the previous example is defined as a child of `PortfolioMessage` which wraps it like this: ```rs pub enum PortfolioMessage { ... // A message that carries the `DocumentMessage` child enum as data #[child] Document(DocumentMessage), ... } ``` Likewise, the `PortfolioMessage` enum is wrapped by the top-level `Message` enum. The dispatcher operates on the queue of these base-level `Message` types. So for example, the `DeleteSelectedLayers` message mentioned previously will look like this as a `Message` data type: ```rs Message::Portfolio( PortfolioMessage::Document( DocumentMessage::DeleteSelectedLayers ) ) ``` Writing out these nested message enum variants would be cumbersome, so that `#[child]` attribute shown earlier invokes a proc macro that automatically implements the `From` trait, letting you write this instead to get a `Message` data type: ```rs DocumentMessage::DeleteSelectedLayers.into() ``` Most often, this is simplified even further because the `.into()` is called for you when pushing a message to the queue with `.add()` or `.add_front()`. So this becomes as simple as: ```rs responses.add(DocumentMessage::DeleteSelectedLayers); ``` The `responses` message queue is composed of `Message` data types, and thanks to this system, child messages like `DocumentMessage::DeleteSelectedLayers` are automatically wrapped in their ancestor enum variants to become a `Message`, saving you from writing the verbose nested form.