Hello,
If I use this technology in an Electron JS desktop application to load a C# AOT or .Net module then on Mac OS (and also Linux) the call to app.quit() hangs and the app does not closes normally, needing to do a force quit from Activity Monitor.
It happens with the simplest possible module found in your examples folder, as it is too much code I have made a demo app here:
[daniacedue/electron-nodeapi] (https://github.com/daniacedue/electron-nodeapi)
As a workaround I have tried brutally closing the process after some time or in the Electron's app.quit() handler by doing process.kill(process.pid, "SIGKILL"); but this is problematic because I need electron-updater npm package to be able to gracefully close in order to kick-off an update.
The problem is quite complex, I have also tried asking some chatbots to figure out what is happening by interpreting some stack traces and spindumps and it has something to do with low-level NAPI calls freezing the main process when doing the cleanup.
I have tried numerous "black box" possible fixes from both C# and JS (like explicit "destroy" function triggering dispose pattern because I was doubting the finalizers being the culpit) but without success.
Thanks!
Hello,
If I use this technology in an Electron JS desktop application to load a C# AOT or .Net module then on Mac OS (and also Linux) the call to
app.quit()hangs and the app does not closes normally, needing to do a force quit from Activity Monitor.It happens with the simplest possible module found in your examples folder, as it is too much code I have made a demo app here:
[daniacedue/electron-nodeapi] (https://github.com/daniacedue/electron-nodeapi)
As a workaround I have tried brutally closing the process after some time or in the Electron's
app.quit()handler by doingprocess.kill(process.pid, "SIGKILL");but this is problematic because I need electron-updater npm package to be able to gracefully close in order to kick-off an update.The problem is quite complex, I have also tried asking some chatbots to figure out what is happening by interpreting some stack traces and spindumps and it has something to do with low-level NAPI calls freezing the main process when doing the cleanup.
I have tried numerous "black box" possible fixes from both C# and JS (like explicit "destroy" function triggering dispose pattern because I was doubting the finalizers being the culpit) but without success.
Thanks!