Merge lp:~brandontschaefer/mir/wait-surface-fullscreen-fix-test into lp:mir

[Brandon Schaefer]

Need to wait for the surfaces to be fullscreen before we do any kind of pointer events!

Need to wait for the surfaces to be fullscreen before we do any kind of pointer events!

[Chris Halse Rogers (raof)]

Eh, sure.

review: Approve

[PS Jenkins bot (ps-jenkins)]

FAILED: Continuous integration, rev:3104
http://jenkins.qa.ubuntu.com/job/mir-ci/5561/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-vivid-i386-build/4888
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-clang-vivid-amd64-build/3795/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-vivid-touch/4831/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-wily-touch/788/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-wily-amd64-ci/1715/console
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/788
        deb: http://jenkins.qa.ubuntu.com/job/mir-wily-i386-ci/788/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4832
        deb: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/4832/artifact/work/output/*zip*/output.zip
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-runner-touch/7410/console
    SUCCESS: http://s-jenkins.ubuntu-ci:8080/job/touch-flash-device/25161
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-wily-armhf/789/console

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/5561/rebuild

review: Needs Fixing (continuous-integration)

[Alan Griffiths (alan-griffiths)]

OK. (Although it isn't obvious that "surface_fullscreen" is needed or has the right cardinality.)

review: Approve

[Brandon Schaefer (brandontschaefer)]

surface_fullscreen isnt 100% needed but it at lease asserts the surfaces are fullscreen before we start the tests.

A better name could be surfaces_become_fullscreen

[Kevin DuBois (kdub)]

change is an improvement, but could there be further races around the visibility of the surfaces? ie, if we changed the server to service the swap buffers after the fullscreen request, the surface could still be hidden and miss the fake input injection. I have a round of fixes to similar situations in other test suites coming around, so lgtm, and I'll revisit that race when proposing those other fixes.

review: Approve

=== modified file 'tests/acceptance-tests/test_surface_raise.cpp'
--- tests/acceptance-tests/test_surface_raise.cpp	2015-11-05 04:13:48 +0000
+++ tests/acceptance-tests/test_surface_raise.cpp	2015-11-17 05:08:50 +0000
@@ -66,6 +66,16 @@
         mir_surface_apply_spec(surface1, spec);
         mir_surface_apply_spec(surface2, spec);
         mir_surface_spec_release(spec);
+        
+        bool surface_fullscreen = mt::spin_wait_for_condition_or_timeout(
+            [this]
+            {
+                return mir_surface_get_state(surface1) == mir_surface_state_fullscreen &&
+                       mir_surface_get_state(surface2) == mir_surface_state_fullscreen;
+            },
+            std::chrono::seconds{max_wait});
+
+        EXPECT_TRUE(surface_fullscreen);
 
         mir_connection_set_lifecycle_event_callback(connection, lifecycle_changed, this);
     }