Closed whot closed 1 month ago
Thanks, I'll try applying this to the packages where I saw #166 and see what happens.
This isn't solving the issue for me. When I run the test in a loop, I'm variously either seeing success, the same failure mode as before, or a segfault inside libportal.
Following up on the segfault is probably the best route, since fixing that might also fix #166.
Following up on the segfault is probably the best route
@smcv I think I got it - this reliably passes on my machine now
Alright, this time it does work. So in the end it was a race condition between when the signal was sent out and when the test registered for the signal. The 500ms simply weren't enough.
Yesterday's patch had a bug - to make it nice python I dropped the result but that had the side-effect of Python cleaning up the XdpSession
object (which was only exposed when I rebased on top of 0f8f627dd86b80f380dd6f0a02b11981a7c8a5e1). So keeping the session
variable alive until the test fixes that particular issue now.
This may or may not fix #166 but ever optimistically we go for the former.
cc @smcv, @GeorgesStavracas