2025.05 Viewer shutdown unresponsiveness

[marchcat]

Not sure why you guys had to change the viewer shutdown again, but if the plan was to get the viewer stuck and becoming unresponsive for 50+% of the time, then it worked great!

Originally posted by @Ansariel in 9117a37

[Ansariel]

Seems to be related to thread queue again: In Firestorm log, I see that the threadsafe queue is running full.

Also: The implementation in 2025.5 looks different from the last know "working"(?) version in former maint-c branch. Shouldn't they be identical?

[marchcat]

The former maint-c also includes the contents of the former (now archived) develop, so these files may differ.
And that is the point of this issue.

[marchcat]

Easy test for the fix:

1. Run the viewer with Local Windows Debugger in Visual Studio;
2. Log in;
3. Close the viewer.

With the fix, it takes 2–4 seconds to close the viewer (and the debug process, respectively); without it, the viewer may hang for over 30 seconds. The results are less dramatic outside of debugging.