When the CLI was first introduced, it would always lazily load the application, and then resolve that lazy import the first time the app was needed. For therun command, in debug mode, that would beafter the server started, so that errors would show up in the debugger and the reloader would work across errors. This meant that errors wouldn't show up immediately when running the command, which was confusing. The--eager-loading/--lazy-loading option controlled whether that was disabled/enabled regardless of debug mode.

Later, this behavior was changed so that the app is always eagerly loaded the first time, and only lazily loaded on reloads. This makes errors show up consistently when running any command, includingrun, while still allowing the reloader to work across errors.

There shouldn't be a reason now to control loading. Errors will always be shown immediately in the terminal when a command is run. Lazy loading should always be used within the reloader to handle errors.