[−][src]Module nvim_rs::examples::scorched_earth

Scorched earth

A minimal port of scorched earth by boxofrox to nvim-rs. Works the same, but foregoes any error handling, removes the customisation of color, and removes some abstractions that aren't helpfull in an example.

Usage

First, build this example via

cargo build --example scorched_earth

The binary will be located in target/debug/examples/scorched_earth.

Follow the steps described for the original plugin. Of course, you don't need to build it. Before step 4, put

let g:scorched_earth_program = '<path to nvim-rs>/target/debug/examples/scorched_earth'

into init.vim. That's it, fire it up and enjoy.

Description

Some overview over the code:

  • The handler struct NeovimHandler needs to contain some plugin state, namely two cursor positions start and end. It needs to be Send and Sync, and we need mutable access, so we wrap it in a Arc<Mutex<_>>. Note that we're using the Mutex from nvim-rs, which is a re-export from Tokio.

  • Implementing the Handler trait requires some magic because of the async functions, we we use the async_trait macro.

  • We use Stdout as the type for the Writer because neovim acts as our parent, so it reads from our stdout. Note that this is the async version re-exported from Tokio.

  • We only implement handle_notify since we don't want to serve requests. It gets a Neovim passed that we can use to send requests to neovim. All requests are async methods, so we need to await them.

  • The main function is denoted #[tokio::main] to use async notation, but it would be perfectly feasible to explicitely create a runtime and use that.

  • After creation of the handler, we connect to neovim via one of the new_* functions. It gives back a Neovim instance which we could use for requests, and a handle for the io loop.

  • The plugin quits by ending the IO task when neovim closes the channel, so we don't need to do anything special. Any cleanup-logic can happen after the IO task has finished. Note that we're loosing access to our Handler, so we might need to implement Drop for it, see the example.

  • After the IO task has finished, we're inspecting the errors to see why it went. A join error simply gets printed, then we inspect potential errors from the io loop itself. First, if we did not see a general reader error, we try to send some last notification to the neovim user. Secondly, we quietly ignore the channel being closed, because this usually means that it was closed by neovim, which isn't always an error.

    Note: A closed channel could still mean an error, so the plugin has the option to react to this.