intermittent acceptance test bug in TestClientInput
Bug #1196744 reported by
Kevin DuBois
This bug affects 4 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Mir |
Fix Released
|
Medium
|
kevin gunn |
Bug Description
bit of a heisenbug, but with current lp:mir, running the TestClientInput.* suite fails jenkins from time to time.
I've poked around this a bit (it happens much more on some of my refactorings of ms/msh) and it appears that the test suite waits in the server side to see the surface, then sends the simulated input.
The client must create a surface and then set a callback for input events, and (I think) in the time between surface creation and input callback registration, the server sometimes sends the input (which is lost because the callback hasn't been registered yet)
Related branches
lp:~kgunn72/mir/change-test-timeout-and-fix-fence
- Alexandros Frantzis (community): Approve
- PS Jenkins bot (community): Approve (continuous-integration)
-
Diff: 135 lines (+26/-12)5 files modifiedinclude/test/mir_test_framework/cross_process_sync.h (+5/-4)
tests/acceptance-tests/test_client_input.cpp (+9/-6)
tests/mir_test_framework/cross_process_sync.cpp (+10/-0)
tests/mir_test_framework/input_testing_server_options.cpp (+1/-1)
tests/mir_test_framework/testing_process_manager.cpp (+1/-1)
Changed in mir: | |
assignee: | nobody → Robert Carr (robertcarr) |
Changed in mir: | |
status: | Confirmed → Fix Released |
no longer affects: | mir (Ubuntu) |
Changed in mir: | |
assignee: | Robert Carr (robertcarr) → kevin gunn (kgunn72) |
To post a comment you must log in.
its a bit of a heisenbug on my desktop, not on my nexus 4! very repeatable with filter= "TestClientInpu t.*" --gtest_repeat=100
./acceptance-tests --gtest_