[tor-bugs] #16311 [Tor Browser]: Fix navigation timing in ESR 38

[Tor Bug Tracker & Wiki]

#16311: Fix navigation timing in ESR 38
-------------------------+-------------------------------------------------
     Reporter:  gk       |      Owner:  gk
         Type:  defect   |     Status:  assigned
     Priority:  normal   |  Milestone:
    Component:  Tor      |    Version:
  Browser                |   Keywords:  ff38-esr, tbb-fingerprinting,
   Resolution:           |  GeorgKoppen201507, tbb-5.0a4,
Actual Points:           |  TorBrowserTeam201508R
       Points:           |  Parent ID:
-------------------------+-------------------------------------------------

Comment (by gk):

 Replying to [comment:13 boklm]:
 > Replying to [comment:11 gk]:
 > > Makes sense but I am wondering why that is not visible when testing
 with 5.0a3 locally.
 >
 > When I try it locally with 5.0a3, it fails.
 >
 > A possible reason is that you ran it with --enable-tests=navigation-
 timing, rather than --enable-tests=tor_bootstrap,navigation-timing. The
 test is accessing network, so it needs the tor daemon to be started
 before. If there is no tor daemon running, there is a proxy error when
 loading the page, and unfortunately it seems that
 controller.waitForPageLoad does not consider this as an error. Apparently
 navigation timing returns only zeros on a "proxy error" page, so the test
 does not fail. To avoid this I will look if we can make the tests fail on
 proxy errors, or make the tor_bootstrap test be run automatically when a
 test that needs it is enabled.

 Yeah, I just ran it with `--enable-tests=navigation-timing`. I think I
 prefer the second option. Otherwise one needs to keep in mind all the
 dependent tests one has to run, too, in order to run a particular test
 which does not scale very well.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/16311#comment:14>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online