$ ./out/Release/compositor_unittests --single-process-tests --gtest_filter=LayerAnimationElementTest.*
...
[ RUN      ] LayerAnimationElementTest.ToString
../../ui/compositor/layer_animation_element_unittest.cc:436: Failure
      Expected: "LayerAnimationElement{name=ThreadedOpacityTransition, id=1, group=42, " "last_progressed_fraction=0.00, start_frame_number=0}"
      Which is: "LayerAnimationElement{name=ThreadedOpacityTransition, id=1, group=42, last_progressed_fraction=0.00, start_frame_number=0}"
To be equal to: element->ToString()
      Which is: "LayerAnimationElement{name=ThreadedOpacityTransition, id=12, group=42, last_progressed_fraction=0.00, start_frame_number=0}"
[  FAILED  ] LayerAnimationElementTest.ToString (0 ms)


The reason is that the test checks the animation_id_, which is set from a global counter that is incremented each time an animation is created. If the tests are running in the same process, this will increment from earlier tests and be wrong for LayerAnimationElementTest.ToString.

Low priority since I believe tests are usually run in multiple-process mode, but I don't think tests should fail in single process mode either :).

More generally I don't understand why we have tests explicitly checking ToString? Walking the callers of LayerAnimationElement::ToString() in codesearch, it looks like this (and LayerAnimationSequence::ElementsToString()) are only used in tests - shouldn't the tests be explicitly checking variable values rather than doing string comparison?