Thanks a lot for your information. I noticed that browser_focus_uitest.cc, which is in interactive ui tests, also uses IN_PROC_BROWSER_TEST and simulates key events (Tab key). I tried to move it into browser_tests and run on trybot, and amazingly it passed. I can't see any real difference between browser_focus_uitest and my test in simulating key events. Regards James Su
2009/9/10 Scott Violet <[email protected]> > To add to what Eric says this is part of the reason for the > interactive ui tests. Interactive ui tests run logged in so that you > can generate key/mouse events and have everything work. I believe Jay > is in the process of splitting these tests up so that we could have > some interactive ui tests that use IN_PROC_BROWSER_TEST. > > -Scott > > On Thu, Sep 10, 2009 at 7:37 AM, Erik Kay<[email protected]> wrote: > > > > Another issue to be aware of is that on the bots, they're running > > without an active desktop. This can lead to a few things behaving > > slightly differently than your local machine. For example, there will > > be no active window, so a call like > > BrowserList::GetLastActiveWithProfile(...) will always return NULL. > > One way you can simulate this behavior locally is to lock your screen > > when you run your test. Another (less reliable) way is to quickly > > give another window focus when you launch your test. > > > > There are other differences as well, maybe others can weigh in with > > more specifics. > > > > Erik > > > > > > On Thu, Sep 10, 2009 at 3:15 AM, James Su <[email protected]> wrote: > >> Hi, > >> Recently I'm working on an automated test of autocomplete edit view, > see > >> CL: http://codereview.chromium.org/177052. It's an in process browser > test. > >> It tests the functionalities of autocomplete edit view (omnibox) by > >> simulating key events. It's supposed to run on both Linux and Windows, > and > >> it runs without any problem on both local Linux machine and Linux > trybot. It > >> also runs on local Windows machine, but it always fails on Windows > trybot. > >> Checking the log, I found that it's timed out when sending key event. > Then > >> I'm wondering if it's a bug or limitation of windows trybot, or I did > >> something wrong? > >> And when I run the test on a local Windows machine, it sometimes fails > >> with following error: > >> [2820:5668:0910/131457:263873000:FATAL:navigation_controller.cc(492)] > Check > >> failed: !GetActiveEntry(). Got an invalid page ID but we seem to be > >> navigated to a valid page. This should be impossible. > >> c:\chromium\src\base/test_suite.h(108): error: Failed > >> [2820:5668:0910/131457:263873000:FATAL:navigation_controller.cc(492)] > Check > >> failed: !GetActiveEntry(). Got an invalid page ID but we seem to be > >> navigated to a valid page. This should be impossible. > >> Backtrace: > >> StackTrace::StackTrace [0x1029B651+33] > >> (c:\chromium\src\base\debug_util_win.cc:226) > >> logging::LogMessage::~LogMessage [0x1025DDFA+618] > >> (c:\chromium\src\base\logging.cc:540) > >> NavigationController::ClassifyNavigation [0x1055B094+180] > >> > (c:\chromium\src\chrome\browser\tab_contents\navigation_controller.cc:494) > >> NavigationController::RendererDidNavigate [0x1055ACD6+198] > >> > (c:\chromium\src\chrome\browser\tab_contents\navigation_controller.cc:424) > >> TabContents::DidNavigate [0x1062DF98+248] > >> (c:\chromium\src\chrome\browser\tab_contents\tab_contents.cc:1953) > >> RenderViewHost::OnMsgNavigate [0x105710EB+459] > >> (c:\chromium\src\chrome\browser\renderer_host\render_view_host.cc:955) > >> ... > >> Is it a bug of chrome or something wrong in my test? > >> Regards > >> James Su > >> > > >> > > > > > > > > > --~--~---------~--~----~------------~-------~--~----~ Chromium Developers mailing list: [email protected] View archives, change email options, or unsubscribe: http://groups.google.com/group/chromium-dev -~----------~----~----~----~------~----~------~--~---
