[Bug]: Calling .center() on BrowserWindow with parent now centers relative to parent instead of screen

[neobyteUK]

Preflight Checklist

• I have read the Contributing Guidelines for this project.
• I agree to follow the Code of Conduct that this project adheres to.
• I have searched the issue tracker for a bug report that matches the one I want to file, without success.

Electron Version

30.0.1

What operating system are you using?

Windows

Operating System Version

Windows 10 Pro 22H2 19045.4291

What arch are you using?

x64

Last Known Working Electron version

27

Expected Behavior

As per documentation for .center() : "Moves window to the center of the screen.".

https://www.electronjs.org/docs/latest/api/browser-window#wincenter

Actual Behavior

When a BrowserWindow has a parent window defined, calling .center() on the child centers the window relative to the parent, not the screen.

Testcase Gist URL

https://gist.github.com/yangannyx/277e71dc348b289cde39e94b4b59d683

Additional Information

This issue does not exist in 27.3.11 but DOES exist in 28.0.0 onwards.

Credit to yangannyx for creating the gist in response to my initial query regarding this on discord.

[yangannyx]

Hello! To add to this, I'm seeing a few differences in behavior between MacOS and Windows on child windows.

1. MacOS child windows get centered relative to the screen, while Windows child windows become centered relative to the parent screen.
2. Dragging the parent window in MacOS will also drag the child windows to preserve their distance/position to the parent window. On Windows, dragging the parent window has no effect on the child window's position.

If it's manageable, I think this would be cool to parameterize in some way

[codebytere]

Looks like this might be an upstream-instigated change, as our Windows implementation hasn't changed in ~6 years:

electron/shell/browser/native_window_views.cc

Line 1073 in 76f7bbb

widget()->CenterWindow(GetSize());

[codebytere]

Good news and bad news: I found the CL that caused this after a quick sprint through Chromium: https://chromium-review.googlesource.com/c/chromium/src/+/4916277, but the CL indicates the previous behavior was unintended and that the new behavior (centering relative to parent) is correct and intended in the method we call. I'll see what the potential options are, as we should ensure the end-user experience remains the same as it was and adheres to expectations from documentation.

[codebytere]

@yangannyx if you could, it'd be helpful to make this:

Dragging the parent window in MacOS will also drag the child windows to preserve their distance/position to the parent window. On Windows, dragging the parent window has no effect on the child window's position.

its own issue!

[yangannyx]

@codebytere, I actually see there's existing electron documentation that describes the drag behavior discrepancy here. I'm leaning towards not opening an issue for this reason, since it doesn't seem like a regression