@theia/remote NPM

Description

This package implements functionality to connect to remote systems using Theia. This facilitates features similar to the features offered by Microsoft's popular Remote-SSH, Dev Containers or WSL extensions for VSCode.

Package Architecture

The following explains the basic flow of any remote connection. It will be exemplified using the remote SSH feature:

When the user runs the SSH: Connect to Host... command, we send the host info to the local backend. The corresponding RemoteSSHConnectionProvider is scoped to the current connection and can request additional information from the user, such as SSH key passphrases.
Once the RemoteSSHConnectionProvider has every information it needs, it creates a SSH connection and registers this connection to the general RemoteConnectionService. Every RemoteConnection type implements an interface that is able to handle 3 kinds of messages to the remote system: 1. Executing commands in the shell of the remote system 2. Copying data to the remote
Once the connection has been established, a setup process takes place on the remote system:
1. Identifying the remote platform (i.e. Windows, MacOS or Linux). This information is needed for all the following steps.
2. Setting up various directories for storing the application and its dependencies.
3. Download and install the correct Node.js version for the remote platform.
4. Packaging, copying, and unpackaging the local backend to the remote backend.
  1. Every Theia extension can register RemoteCopyContribution binding to copy certain files from the current system. This contribution point is used for files that are used in all operating systems.
  2. They can also register RemoteNativeDependencyContribution bindings to download and copy native dependencies for the remote system. The downloaded files are on a per-platform basis.
5. Using the node version that was installed in step 3, we now start the main.js of the backend application. We start the backend with --port=0, so that it searches for any available port. It will print the port to the console. The setup either returns with a setup error or the port of the remote server on the remote system.
With the remote server/port in place, the backend sets up a local proxy server on a random port. It instructs the RemoteConnection object to forward any HTTP request to this proxy server to the remote server.
The backend will return from the initial request from (1) with a new local proxy port. The frontend sets this port in the url and reload itself.
The frontend is now connected to the remote backend by connecting to the local proxy port.
The frontend now performs its normal messaging lifecycle, establishing connections to backend services. Although these backend services live on a different remote system, the frontend handles them as if they belong to the local backend.