(No comment was entered for this change.)

The problem is with AmIBundled, see http://groups.google.com/a/chromium.org/group/chromium-dev/browse_thread/thread/f3500e4fcc468030#

(No comment was entered for this change.)

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=57088 

------------------------------------------------------------------------
r57088 | rohitrao@chromium.org | 2010-08-23 12:09:28 -0700 (Mon, 23 Aug 2010) | 7 lines
Changed paths:
   M http://src.chromium.org/viewvc/chrome/trunk/src/base/mac_util.mm?r1=57088&r2=57087

[Mac] Add diagnostic logging to AmIBundled() to help with bot failures.

This change should be reverted as soon as we figure out what's going wrong.

BUG=52918
TEST=None
Review URL: http://codereview.chromium.org/3197012
------------------------------------------------------------------------

Partial revert is ready in http://codereview.chromium.org/3164035/show

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=57194 

------------------------------------------------------------------------
r57194 | rohitrao@chromium.org | 2010-08-24 09:45:45 -0700 (Tue, 24 Aug 2010) | 5 lines
Changed paths:
   M http://src.chromium.org/viewvc/chrome/trunk/src/base/mac_util.mm?r1=57194&r2=57193

[Mac] Revert the diagnostic messages added in r57088.

BUG=52918
TEST=Less log spam.
Review URL: http://codereview.chromium.org/3164035
------------------------------------------------------------------------

Ok, so there's what is happening on Mac.  The FILE_EXE is getting overridden, and that then cause DIR_EXE to get recalculated.  Then when DIR_SOURCE_ROOT is calculated, but it apples the logic based on the running executable to the overridden DIR_EXE.  Causing the wrong value.  I'll build up a CL to fix that, but there maybe or upstream things that get confused.

As far as I can tell, Windows/Linux get lucky here because the overrides are always in the same directory, the fact that Mac bundlers are directories trees is what chokes things here.

yea, looking at chrome_paths and chrome_paths_mac, the dump of the cache is likely to cause more problems upstream for you.

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=57237 

------------------------------------------------------------------------
r57237 | thomasvl@chromium.org | 2010-08-24 14:32:35 -0700 (Tue, 24 Aug 2010) | 5 lines
Changed paths:
   M http://src.chromium.org/viewvc/chrome/trunk/src/base/base_paths_mac.mm?r1=57237&r2=57236

[Mac] if someone overrides FILE_EXE/DIR_EXE in the path service, it can cause the calculation for DIR_SOURCE_ROOT to go wrong.  Avoid this by not using the value from the path service.

BUG=52918
TEST=green tree
Review URL: http://codereview.chromium.org/3176036
------------------------------------------------------------------------

(No comment was entered for this change.)

(No comment was entered for this change.)

Verified label updated by AutoAllocator, contact AmolK or KrisR for details

When you patch in http://codereview.chromium.org/2805100/show locally, there problem is still there. This shouldn't rely on the caching implementation inside PathService.

The real problem is people are changing FILE_EXE and then the app can't find things.  the real fix would be to figure out why FILE_EXE is getting overridden and correct that behavior.

ps- you probably need to update the summary if you are morphing this into that larger problem.  Windows/Linux just works by luck (they have is changed to something within the same directory, so DIR_EXE doesn't get changed). If we moved where a binary went, those platform would also blow up for the same reason.

This is fixed now. Thanks, Thomas!

Verified label updated by AutoAllocator, contact AmolK or KrisR for details

(No comment was entered for this change.)

I see some FLAKY test failures like http://build.chromium.org/buildbot/waterfall/builders/Mac10.5%20Tests%20(dbg)(3)/builds/10939/steps/browser_tests/logs/stdio, where we're looking for testserver.py in the wrong place:

Note: Google Test filter = SSLUITest.TestHTTP
[==========] Running 1 test from 1 test case.
[----------] Global test environment set-up.
[----------] 1 test from SSLUITest
[ RUN      ] SSLUITest.TestHTTP
[30866:523:0820/151022:2246034617242922:ERROR:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/first_run/first_run_mac.mm(63)] Not implemented reached in static bool FirstRun::ProcessMasterPreferences(const FilePath&, FirstRun::MasterPrefs*)
[30866:23299:0820/151022:2246034621664602:INFO:/b/slave/chromium-dbg-mac-builder/build/src/net/base/host_resolver_impl.cc(1101)] IPv6Probe forced AddressFamily setting to ADDRESS_FAMILY_IPV4
[30866:523:0820/151022:2246034629245920:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/extensions/extensions_service.cc(686)] Sending EXTENSION_LOADED
[30866:23299:0820/151022:2246034632405780:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/net/predictor_api.cc(372)] DNS Prefetch service started
[30866:523:0820/151022:2246034651711683:INFO:/b/slave/chromium-dbg-mac-builder/build/src/webkit/support/../glue/plugins/plugin_list.cc(217)] Loaded plugin list in 19 ms.
[30866:21763:0820/151022:2246034728212826:WARNING:/b/slave/chromium-dbg-mac-builder/build/src/net/base/file_stream_posix.cc(350)] Failed to open file: 2 (/private/var/folders/Tf/TfOs8gm+HkuQrkICa3LquE+++TM/-Tmp-/.org.chromium.3Nzx4o/test_user_data/Default/Last Tabs)
[30866:523:0820/151022:2246034736330510:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/sync/profile_sync_service.cc(104)] Unofficial build, using dev channel sync server.
[30866:523:0820/151022:2246034736467700:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/sync/profile_sync_service.cc(130)] Starting ProfileSyncService.
[30866:523:0820/151022:2246034736496686:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/sync/profile_sync_service.cc(212)] Using https://clients4.google.com/chrome-sync/dev for sync server URL.
[30866:523:0820/151022:2246034736767052:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/sync/profile_sync_service.cc(374)] Clearing Sync DB.
[30866:523:0820/151023:2246034898205794:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/history/history.cc(749)] History backend finished loading
[30866:21763:0820/151023:2246034906222059:INFO:/b/slave/chromium-dbg-mac-builder/build/src/chrome/common/important_file_writer.cc(71)] successfully saved /private/var/folders/Tf/TfOs8gm+HkuQrkICa3LquE+++TM/-Tmp-/.org.chromium.3Nzx4o/test_user_data/Default/Bookmarks
/System/Library/Frameworks/Python.framework/Versions/2.5/Resources/Python.app/Contents/MacOS/Python: can't open file '/b/slave/net/tools/testserver/testserver.py': [Errno 2] No such file or directory
[30866:523:0820/151033:2246045383398036:ERROR:/b/slave/chromium-dbg-mac-builder/build/src/net/test/test_server.cc(251)] Failed to connect to server
/b/slave/chromium-dbg-mac-builder/build/src/chrome/browser/ssl/ssl_browser_tests.cc:124: Failure
Value of: test_server()->Start()
  Actual: false
Expected: true
[  FAILED  ] SSLUITest.TestHTTP (11115 ms)

In test_server.cc, we have:


  FilePath testserver_path;
  if (!PathService::Get(base::DIR_SOURCE_ROOT, &testserver_path)) {
    LOG(ERROR) << "Failed to get DIR_SOURCE_ROOT";
    return false;
  }
  testserver_path = testserver_path
      .Append(FILE_PATH_LITERAL("net"))
      .Append(FILE_PATH_LITERAL("tools"))
      .Append(FILE_PATH_LITERAL("testserver"))
      .Append(FILE_PATH_LITERAL("testserver.py"));

so somehow we got DIR_SOURCE_ROOT = /b/slave instead of /b/slave/chromium-dbg-mac5-tests-3/build/src.

This issue has been closed for some time. No one will pay attention to new comments.
If you are seeing this bug or have new data, please click New Issue to start a new bug.

(No comment was entered for this change.)