input_pipeline > Integration

Reviewed on: 2022-03-22

This document describes how to integrate the input pipeline within a larger Rust program. For example, the input pipeline is integrated with Scene Manager for some builds.

Requirements

The code that deals with the input pipeline must run on a LocalExecutor. For existing code, this is already true. SceneManager uses fuchsia_async::run_singlethreaded, so it uses a LocalExecutor.

Alternatives

A program might want to isolate itself from the risk that a bug in the input pipeline library (e.g. infinite loop) will keep the rest of the program from making progress. Such programs can

1. Spawn a dedicated thread to run a LocalExecutor which creates and runs the input pipeline, OR
2. Run the input pipeline as a standalone component, and communicate with the pipeline over FIDL.

Component Manifest

The library provides a client shard to ensure proper routing is available. This shard can be included in the integrating component's manifest file.

Structured Configuration

The input pipeline uses structured configuration by declaring a config value in the integrating component's manifest, such as in //src/ui/bin/scene_manager/meta/scene_manager.cml:

config: {
    supported_input_devices: {
        type: "vector",
        element: {
            type: "string",
            max_size: 12,
        },
        max_count: 6,
    },
},

This value can also be set in product assembly via the following configuration schema:

product_assembly_configuration("my_product") {
  platform = {
    input = {
      supported_input_devices = [
        "button",
        "touchscreen",
      ]
    }
  }
}

It can also be set to a default value such as for testing:

fuchsia_structured_config_values("test_config") {
  cm_label = ":manifest"
  values = {
    supported_input_devices = [
      "button",
      "keyboard",
      "lightsensor",
      "mouse",
      "touchscreen",
    ]
  }
}