[Bug]: [6.1.7 - 7.0.10] repaint issue: InvalidateRect() called with wrong rect size if coming from keyboard hook #1366
Comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
{{ message }}
Detailed steps on how to reproduce the bug
We have the issue that our plugin has repaint issues if a juce::List is repainted due to key up/down presses, selecting a row via mouse is fine. It only happens in Bitwig VST2/VST3 and a Windows 10/11 DPI scale of 125%.
It was quite some work to track this down, here is the result:
The reason for this is that the function
static RECT getWindowClientRect (HWND hwnd)returns a different window size if its called from thekeyboardHookCallback()compared to being called from thewindowProc()Example result when called from the keyboardHookCallback:
GetWindowRect HWND=00590fcc 88 1977 3296 738 MapWindowPoints HWND=00590fcc 0 0 1319 650Example result when called from the windowProc:
GetWindowRect HWND=00590fcc 110 1991 3640 923 MapWindowPoints HWND=00590fcc 0 0 1649 813If we do the math, we can see that the reported window size for the keyboard hook compared to the windowProc is 100/125 too small, exactly the inverse of the DPI setting.
This causes all
internalRepaint()calls to report a downscaled rectangle to the win32 functionInvalidateRect(). As Win32 responds with WM_PAINT messages via windowProc (where the reported window size is correct again), the redraw regions are all messed up.Unfortunately I do not know enough about Win32 programming to provide a fix on my own, I can only assume that Windows is not aware of the current DPI setting when being called from the keyboard hook.
Hope this helps.
What is the expected behaviour?
The invalidated regions should have the correct positions & sizes, regardless of any DPI setting
Operating systems
Windows
What versions of the operating systems?
Tested on Windows 11 & Windows 10
Architectures
x86_64
Stacktrace
Plug-in formats (if applicable)
VST2, VST3
Plug-in host applications (DAWs) (if applicable)
Bitwig Studio 5.1.3
Testing on the
developbranchThe bug is present on the
developbranchCode of Conduct
The text was updated successfully, but these errors were encountered: