There are a couple non-trivial cases that would be nice to test for.
Created attachment 46296 [details]
There is one failing subtest. The issue is minor, I don't even want to file a bug about it yet.
Committed revision 53095.
This new test fails on the Qt bot:
Qt just needs expected failure results checked in (and maybe a bug to track fixing platform-specific problems).
Can you please do so? The bots were green before your and Darin's checkins.