If the entire gtest run crashes/failes/times out, the tests that
were not run are marked as 'unknown'. These tests are reported
by GetAllBroken() and are retried by BaseTestSharder.
Get rid of overall_* flags which are now redundant.
BUG=171466
Review URL: https://codereview.chromium.org/11896050
git-svn-id: http://src.chromium.org/svn/trunk/src/build@178345 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
Various chrome test suites use the infrastructure in net::LocalTestServer and net/tools/testserver.py to create local test server instances against which to run automated tests. Sync tests use reference implementations of sync and xmpp servers, which build on the testserver infrastructure in net/. In the past, the sync testserver was small enough that it made sense for it to be a part of the testserver in net/. This, however, resulted in an unwanted dependency from net/ onto sync/, due to the sync proto modules needed to run a python sync server. Now that the sync testserver has grown considerably in scope, it is time to separate it out from net/ while reusing base testserver code, and eliminate the dependency from net/ onto sync/. This work also provides us with the opportunity to remove a whole bunch of dead pyauto sync test code in chrome/test/functional.
This patch does the following:
- Moves the native class LocalSyncTestServer from net/test/ to sync/test/.
- Moves chromiumsync{_test}.py and xmppserver{_test}.py from net/tools/testserver/ to sync/tools/testserver/.
- Removes all sync server specific code from net/.
- Adds a new sync_testserver.py runner script for the python sync test.
- Moves some base classes from testserver.py to testserver_base.py so they can be reused by sync_testserver.py.
- Audits all the python imports in testserver.py, testserver_base.py and sync_testserver.py to make sure there are no unnecessary / missing imports.
- Adds a new run_sync_testserver runner executable to launch a sync testserver.
- Removes a couple of static methods from LocalTestServer, that were being used by run_testserver, and refactors run_sync_testserver to use their non-static versions.
- Adds the ability to run both chromiumsync_test.py and xmppserver_test.py from run_sync_testserver.
- Fixes chromiumsync.py to undo / rectify some older changes that broke tests in chromiumsync_test.py.
- Adds a new test target called test_support_sync_testserver to sync.gyp.
- Removes the hacky dependency on sync_proto from net.gyp:net_test_support.
- Updates various gyp files across chrome to use the new sync testserver target.
- Audits dependencies of net_test_support, run_testserver, and the newly added targets.
- Fixes the android chrome testserver spawner script to account for the above changes.
- Removes all mentions of TYPE_SYNC from the pyauto TestServer shim.
- Deletes all (deprecated) pyauto sync tests. (They had all become broken over time, gotten disabled, and were all redundant due to their equivalent sync integration tests.)
- Removes all sync related pyauto hooks from TestingAutomationProvider, since they are no longer going to be used.
- Takes care of a TODO in safe_browser_testserver.py to remove an unnecessary code block.
Note: A majority of the bugs listed below are for individual pyauto sync tests. Deleting the sync pyauto test script fixes all these bugs in one fell swoop.
TBR=mattm@chromium.org
BUG=117559, 119403, 159731, 15016, 80329, 49378, 87642, 86949, 88679, 104227, 88593, 124913
TEST=run_testserver, run_sync_testserver, sync_integration_tests, sync_performance_tests. All chrome tests that use a testserver should continue to work.
Review URL: https://codereview.chromium.org/11971025
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177864 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
This bot is supposed to be a canary for the main waterfall,
but it's configuration got out of sync. The android testers
on main waterfall are running a variety of android
instrumentation tests, but this bot was only running unit
tests. As a result, the latest webkit deps roll broke the
android testers on main waterfall.
Make this bot a 'copied' configuration to keep the set of tests
in sync from now on.
TBR=yfriedman
Review URL: https://codereview.chromium.org/11938011
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177622 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
The Galaxy S3's build of android is of type 'user' (as one would
expect for a production device). This prevents the use of 'adb root',
even on rooted devices, however one can root an S3 and install an
'su' command on it without changing the android build. The changes
in this commit switch to using the 'su' command in place of 'adb root'.
Since writing to a protected file using su is quite complex they
encapsulate this in a the function SetProtectedFileContents()
and make sure that all calls in the Android test framework that write
to protected files call this.
Also, allow variable wait for logcat entries. This is now needed since
other test changes mean that the 10 second default may not be adequate
in some cases.
BUG=169001,163336
Review URL: https://chromiumcodereview.appspot.com/11876021
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177508 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
In preparation to moving release tester to main waterfall,
add zip_step to release builder.
Also adding release trybot configs at same time.
Some configs are implicit by using bot_id substring match.
BUG=164995,167393
TBR=yaron
NOTRY=True
Review URL: https://chromiumcodereview.appspot.com/11975033
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177357 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
We want certain Android specific files, like findbugs_known_bugs.txt, to only have deletions (not even modifications to existing lines). The presubmit check added in this CL checks that a certain list of files only have deletions.
BUG=165608
NOTRY=True
Review URL: https://chromiumcodereview.appspot.com/11583028
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177241 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
This removes unneeded gyp defines from envsetup. I still left the environment variables because some scripts (e.g. findbugs) still depend on them. I'll remove those usages and the environment variables separately.
Review URL: https://codereview.chromium.org/11886053
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177169 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
The benchmarks first print results to stdout, then information about skipped /
failed page to stderr. The test runners also redirect both streams to the same
file. If stdout is not properly flushed, the stderr output might appear in the
middle of the stdout output, and this makes the log parser fail.
R=bulach
BUG=none
Review URL: https://codereview.chromium.org/11886071
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177162 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
The test code was incorrectly counting the number of CPUs, and hence was only
setting performance mode on, at most, n-1 of them. This meant that the
results of the perf tests (at least the startup tests) depended on how Chrome
distributed itself between CPUs.
BUG=163336
Review URL: https://chromiumcodereview.appspot.com/11938002
git-svn-id: http://src.chromium.org/svn/trunk/src/build@177143 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
This is the client-side change for uploading the following
instrumentation test types:
androidwebview_instrumentation_tests
chromiumtestshell_instrumentation_tests
contentshell_instrumentation_tests
Actuall uploading of results will first be done
on a staging server.
BUG=150801
Review URL: https://codereview.chromium.org/11885010
git-svn-id: http://src.chromium.org/svn/trunk/src/build@176758 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
The code was making a call to android_commands.py, which
relied on a copy of adb existing in the path. The CQ
bot does not have third_party/android_tools, and in general
it seems advantageous to not require a full checkout for
syntax checks.
NOTRY=True
Review URL: https://chromiumcodereview.appspot.com/11883019
git-svn-id: http://src.chromium.org/svn/trunk/src/build@176723 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
This should move the last dependence on adb out of buildbot_functions. Also includes various other android buildbot fixes:
- Start logcat monitor before reboot command to catch JB crash.
- Rename bb_tests to bb_device_steps.
Review URL: https://codereview.chromium.org/11857014
git-svn-id: http://src.chromium.org/svn/trunk/src/build@176651 4ff67af0-8c30-449e-8e8b-ad334ec8d88c
boliu@chromium.org
frankf@chromium.org
yusufo@chromium.org
jochen@chromium.org
rsesek@chromium.org
yfriedman@chromium.org
ppi@chromium.org
sbc@chromium.org
bulach@chromium.org
rsimha@chromium.org
michaelbai@chromium.org
ilevy@chromium.org
aberent@chromium.org
sivachandra@chromium.org
peter@chromium.org
marja@chromium.org
skyostil@chromium.org
cjhopman@chromium.org
frankf@google.com
kristianm@chromium.org
viettrungluu@chromium.org
eugenis@chromium.org