Internal error in node-inspect. Please report this bug.

[LEQADA]

Description

When the port is busy – node-inspect throws a timeout exception which is logged as an internal error.

• Version: v9.7.1
• Platform: Windows 10 x64
• Subsystem: node-inspect

node debug .
(node:14108) [DEP0068] DeprecationWarning: `node debug` is deprecated. Please use `node inspect` instead.
There was an internal error in node-inspect. Please report this bug.
Timeout (2000) waiting for 127.0.0.1:9229 to be free
Error: Timeout (2000) waiting for 127.0.0.1:9229 to be free
    at Timeout.setTimeout [as _onTimeout] (node-inspect/lib/_inspect.js:54:14)
    at ontimeout (timers.js:466:11)
    at tryOnTimeout (timers.js:304:5)
    at Timer.listOnTimeout (timers.js:267:5)

Overview

Timeout message is generated here

node-inspect/lib/_inspect.js

Lines 52 to 56 in 3c9073f

	setTimeout(() => {
	didTimeOut = true;
	reject(new Error(
	`Timeout (${timeout}) waiting for ${host}:${port} to be free`));
	}, timeout);

and then catched by a handler

node-inspect/lib/_inspect.js

Lines 348 to 355 in 3c9073f

	function handleUnexpectedError(e) {
	console.error('There was an internal error in node-inspect. ' +
	'Please report this bug.');
	console.error(e.message);
	console.error(e.stack);
	if (inspector.child) inspector.child.kill();
	process.exit(1);
	}

[jkrems]

Suggestion by @bnoordhuis:

Would logging a message followed by process.exit(1) in the timer callback work?

Suggestion by @jkrems:

Alternatively: making it a non-fatal error. "Failed to start app. Please check that the port isn't used by anything else. Use run to try again." etc.. Another solution would be to use a custom code and/or error class (StartupError) and exclude that from the internal error message. But that seems a bit more hacky.

[LEQADA]

@bnoordhuis I think it will work. But I'm not sure is it good to process.exit(1) in 2 places 🤔
@jkrems I'm not quite sure what you mean by making it non-fatal error. To not reject? Create another type of handler? Could you give me an example?

[jkrems]

By "non-fatal error" I meant starting the repl instead of exiting the node-inspect process. Then the user just needs to type restart to retry if they fixed the other process. That might also address cases where the app crashed, the user is trying to run it again (run<enter>), and they run into the same problem (another process is blocking the port).

[jkrems]

But I'd be more than happy to merge a simpler solution that just hides away the "internal error" message somehow. There's always time to improve the UX even further in the future.

[jkrems]

Potential fix is here: #62