NetworkIconTest.GetCellularUninitializedMsg_NoUninitializedMessageExpected is flaky |
||
Issue descriptionToT r519699 on linux desktop args.gn: dcheck_always_on = true is_debug = false is_component_build = true target_os = "chromeos" use_goma = true enable_nacl = false testing/xvfb.py out/Default/ash_unittests --test-launcher-retry-limit=0 --gtest_shuffle --gtest_repeat=100 You'll see several of these failures: [ RUN ] NetworkIconTest.GetCellularUninitializedMsg_NoUninitializedMessageExpected ../../ash/system/network/network_icon_unittest.cc:181: Failure Expected: 0 To be equal to: GetCellularUninitializedMsg() Which is: 23753 [ FAILED ] NetworkIconTest.GetCellularUninitializedMsg_NoUninitializedMessageExpected (1 ms) I see static variables in GetCellularUninitializedMessage(), so it's possible state is leaking from one test to another: https://cs.chromium.org/chromium/src/ash/system/network/network_icon.cc?sq=package:chromium&l=888 It would be good not to use statics there. Kyle, can you take a look? Steven/Leslie, other ideas?
,
Nov 28 2017
Let's move the debouncing to NetworkStateHandler. The correct solution I think is to have handler->GetTechnologyState(Cellualr) return "Enabling" if the state is "Enabled" but there is no Celular DeviceState, and to have GetCellularUninitializedMessage() treat 'initializing' and 'enabling' the same as 'uninitialized'. Does that make sense?
,
Nov 28 2017
Sure, that makes sense, Steven. Thanks! |
||
►
Sign in to add a comment |
||
Comment 1 Deleted