electron: Windows: WorkArea property not updated

The PR was merged into Chromium last night: https://codereview.chromium.org/2829693003/ note this is a Windows-only fix.

It looks like Chromium does not track changes to screen work areas properly, and returns the stale values still to Electron. I opened a Chromium bug: https://bugs.chromium.org/p/chromium/issues/detail?id=712871

and have submitted a PR for a fix that solves the problem for me: https://codereview.chromium.org/2829693003

@z-ax I looked around but was unable to find a related Chromium issue either. I’m surprised this has been an issue for 5 months without something ticketed on their side… Is there someone with more knowledge of Electron’s interface with Chromium that could file an issue related to this and get more visibility on this?

Checked with v1.3.0 - still an issue.

@llhupp Agree, haven’t found any Chromium issue and sadly don’t know the details of Electron APIs good enough to write a proper issue.

@zcbenz Can you write a Chromium issue or link to existing one for this?

@Ciantic Thanks! This is what I was thinking of doing as well. Given up hope that this will be solved any time soon and can’t upgrade Electron until this has been fixed. The only missing part is to know when to get the new params (like display changes). I think there was an issue with WorkArea changes not being notified about as well. But for me this should be enough so thanks.

Here is the code to get it using Node-FFI, using WinAPI crap, it’s ugly as hell.

Note3 I think the X, Y, are now correct, however I think there is some flakiness with below method:

import { app, screen } from "electron";
import * as ffi from "ffi";
import * as ref from "ref";
import * as struct from "ref-struct";

let Point = struct({
    'x' : 'long',
    'y' : 'long',
});
let Rect = struct({
    'left' : 'long',
    'top' : 'long',
    'right' : 'long',
    'bottom' : 'long',
});
const User32 = ffi.Library("User32.dll", {
    'MonitorFromPoint' : ['uint', [Point, 'int']], // point, dword -> hmonitor
    'GetMonitorInfoW' : ['bool', ['uint', 'pointer']] // hmonitor, lpmi -> bool
});
const Shcore = ffi.Library("Shcore.dll", {
    'GetDpiForMonitor' : ['int', ['uint', 'int', 'pointer', 'pointer']] // hmonitor, MONITOR_DPI_TYPE, puint, puint -> HRESULT
});
let getWorkArea = (point: {x: number, y: number}) => {
    let pointInstance = new Point(point);
    let hmonitor = User32.MonitorFromPoint(pointInstance, 2); // MONITOR_DEFAULTTONEAREST == 2
    let dpiXptr = ref.alloc('int');
    let dpiYptr = ref.alloc('int');
    Shcore.GetDpiForMonitor(hmonitor, 0, dpiXptr, dpiYptr);
    let dpiX =  dpiXptr.readInt32LE(0);
    let dpiY = dpiYptr.readInt32LE(0);
    let MonitorInfo = struct({
        cbSize : 'int',
        rcMonitor : Rect,
        rcWork : Rect,
        dwFlags : 'int',
    });
    let monitorInfoPtr = new MonitorInfo({
        cbSize : MonitorInfo.size,
        rcMonitor : new Rect(),
        rcWork : new Rect(),
        dwFlags : 1, // TODO: Force primary display!
    });
    User32.GetMonitorInfoW(hmonitor, monitorInfoPtr.ref());
    let x = monitorInfoPtr.rcWork.left * 96 / dpiX;
    let y = monitorInfoPtr.rcWork.top * 96 / dpiY;
    return {
        x : x > 0 ? Math.ceil(x) : Math.floor(x),
        y : y > 0 ? Math.ceil(y) : Math.floor(y),
        width : Math.ceil((monitorInfoPtr.rcWork.right - monitorInfoPtr.rcWork.left) * 96 / dpiX),
        height : Math.ceil((monitorInfoPtr.rcWork.bottom - monitorInfoPtr.rcWork.top) * 96 / dpiY)
    };
}
app.on('ready', () => {
    let display = screen.getDisplayNearestPoint({
        x : 0,
        y : 0
    });
    let workArea = getWorkArea({
        x : 0,
        y : 0
    });
    console.log("display?", display.workArea);
    console.log("workarea winapi?", workArea);
    app.quit();
});

// On my computer it prints out:
// display? { x: 0, y: 0, width: 2132, height: 1235 }
// workarea winapi? { x: 0, y: 0, width: 2132, height: 1235 }

Is there an issue in Chromium for this regression? Didn’t find.