Can we talk about child_process.fork() and exit codes?

[Carey Williams]

It's been mentioned on this group a few times before, and there are some open tickets:

https://github.com/nwjs/nw.js/issues/4875

https://github.com/nwjs/nw.js/issues/6916

But seemingly not a lot of traction on the issues.

From what I understand from reading around and doing my own experiments: fork is implemented by re-running a new instance nwjs, but presence of a .js file on the command causes it to switch to a "node-like" mode, which behaves like a typical node invocation. So far so good.

But here's where the main problem arises: the script status code is ignored completely.

Take the simple example below:

error.js, contains:

   throw new Error("Bad Exit!")

Running with Node

node error.js

   node error.js    

   /Users/carey/Documents/Work/Github/ChoiceScriptIDE/main/error.js:1

   throw new Error("Bad Exit!")

Error: Bad Exit!

echo $?

127

Running with NWJS Helper (Renderer) -- as with .fork

..Frameworks/nwjs Framework.framework/Versions/97.0.4692.71/Helpers/nwjs Helper (Renderer).app/Contents/MacOS/nwjs Helper (Renderer) error.js

$?

0

The nwjs process doesn't propagate script errors correctly, as long it starts up a valid .JS file, it can do whatever it likes and there's no way to get feedback about that.

It's a little better if the script doesn't exist. From the command line you get an unhelpful error: [0122/150414.532056:FATAL:chrome_main_delegate.cc(1060)] Check failed: command_line.HasSwitch(switches::kProcessType) && !process_type.empty(). Helper application requires --type.

And exit code 133. Unfortunately, from fork itself, this results in null.

I don't know the NWJS codebase, or how it fits together with V8 and Chromium, but I would hope this is something that isn't too difficult to fix. I'd be happy to contribute, as best I can, but I don't really know where to start looking. node.cc and node.js maybe? Or is the whole problem a lot deeper rooted than I'm thinking?

[Benoit Schmitlin]

Yes would love to see some traction as well, pretty sure it affects Windows as well (or something causing a similar 100% CPU symptom at least) ...