Remove serial test support from test-webkitpy.

Remove serial test support from test-webkitpy.

Previously test-webkitpy had a way to annotate tests that could not
be run in parallel with other tests. This was useful in rare occasions,
but such tests usually are flaky and slow, so they were gradually removed
and/or reworked. Ultimately this feature wasn't really being used and
was just a source of needless complexity, so I'm killing it.

R=eseidel@chromium.org
BUG=402172
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=180217

[Dirk Pranke, 2014-08-09 00:00:11]

https://codereview.chromium.org/458643003/diff/1/Tools/Scripts/webkitpy/layou...
File Tools/Scripts/webkitpy/layout_tests/port/android_unittest.py (left):

https://codereview.chromium.org/458643003/diff/1/Tools/Scripts/webkitpy/layou...
Tools/Scripts/webkitpy/layout_tests/port/android_unittest.py:127: def
serial_test_adb_command_path(self):
This test has unpleasant side effects and is testing a code path that

a) is already covered any time we create an AndroidCommands() object

b) shouldn't exist at all (using a static mutable value to cache the path to adb
is just wrong)

c) was originally intended to look for adb in multiple places, which we don't
actually use now.

At some point I'll clean up the other code, but this test does more harm than
good.

[eseidel, 2014-08-09 00:12:14]

Note that the new infra test runner has similar support:
https://chromium.googlesource.com/infra/infra/+/master/test.py

[eseidel, 2014-08-09 00:15:05]

lgtm

[Dirk Pranke, 2014-08-09 00:45:30]

On 2014/08/09 00:12:14, eseidel wrote:
> Note that the new infra test runner has similar support:
> https://chromium.googlesource.com/infra/infra/+/master/test.py

Interesting, I was not aware of that tool. iannucci and I had talked about this
concept in the past, so maybe this is their stab at it (I had even pointed him
at an early version of typ).

There's probably not a good reason to have two tools, so I'll take a look at it
in more detail  and talk to them about the overlap ...

Thanks for the pointer!

[Dirk Pranke, 2014-08-09 01:07:45]

The CQ bit was checked by dpranke@chromium.org

[commit-bot: I haz the power, 2014-08-09 01:08:39]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/dpranke@chromium.org/458643003/1

[commit-bot: I haz the power, 2014-08-09 01:08:45]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-08-09 01:08:46]

Failed to apply patch for Tools/Scripts/webkitpy/test/main.py:
While running patch -p1 --forward --force --no-backup-if-mismatch;
  patching file Tools/Scripts/webkitpy/test/main.py
  Hunk #1 succeeded at 167 (offset 2 lines).
  Hunk #2 FAILED at 193.
  1 out of 2 hunks FAILED -- saving rejects to file
Tools/Scripts/webkitpy/test/main.py.rej

Patch:       Tools/Scripts/webkitpy/test/main.py
Index: Tools/Scripts/webkitpy/test/main.py
diff --git a/Tools/Scripts/webkitpy/test/main.py
b/Tools/Scripts/webkitpy/test/main.py
index
d5db7835fbca263d835d8b5451f366eb83da6c1d..c9d68a3f026b39b1017f82e987f2798ceb787103
100644
--- a/Tools/Scripts/webkitpy/test/main.py
+++ b/Tools/Scripts/webkitpy/test/main.py
@@ -165,14 +165,13 @@ class Tester(object):
 
         self.printer.write_update("Finding the individual test methods ...")
         loader = _Loader()
-        parallel_tests, serial_tests = self._test_names(loader, names)
+        parallel_tests = self._test_names(loader, names)
 
         self.printer.write_update("Running the tests ...")
-        self.printer.num_tests = len(parallel_tests) + len(serial_tests)
+        self.printer.num_tests = len(parallel_tests)
         start = time.time()
         test_runner = Runner(self.printer, loader, self.webkit_finder)
         test_runner.run(parallel_tests, self._options.child_processes)
-        test_runner.run(serial_tests, 1)
 
         self.printer.print_result(time.time() - start)
 
@@ -194,24 +193,13 @@ class Tester(object):
 
     def _test_names(self, loader, names):
         parallel_test_method_prefixes = ['test_']
-        serial_test_method_prefixes = ['serial_test_']
 
         parallel_tests = []
         loader.test_method_prefixes = parallel_test_method_prefixes
         for name in names:
            
parallel_tests.extend(self._all_test_names(loader.loadTestsFromName(name,
None)))
 
-        serial_tests = []
-        loader.test_method_prefixes = serial_test_method_prefixes
-        for name in names:
-           
serial_tests.extend(self._all_test_names(loader.loadTestsFromName(name, None)))
-
-        # loader.loadTestsFromName() will not verify that names begin with one
of the test_method_prefixes
-        # if the names were explicitly provided (e.g., MainTest.test_basic), so
this means that any individual
-        # tests will be included in both parallel_tests and serial_tests, and
we need to de-dup them.
-        serial_tests = list(set(serial_tests).difference(set(parallel_tests)))
-
-        return (parallel_tests, serial_tests)
+        return parallel_tests
 
     def _all_test_names(self, suite):
         names = []

[Dirk Pranke, 2014-08-09 01:15:27]

The CQ bit was checked by dpranke@chromium.org

[commit-bot: I haz the power, 2014-08-09 01:16:35]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/dpranke@chromium.org/458643003/20001

[commit-bot: I haz the power, 2014-08-09 02:16:25]

FYI, CQ is re-trying this CL (attempt #1).
The failing builders are:
  mac_blink_rel on tryserver.blink
(http://build.chromium.org/p/tryserver.blink/builders/mac_blink_rel/builds/19103)

[commit-bot: I haz the power, 2014-08-09 02:22:50]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-08-09 02:22:52]

Try jobs failed on following builders:
  mac_blink_rel on tryserver.blink
(http://build.chromium.org/p/tryserver.blink/builders/mac_blink_rel/builds/19106)

[iannucci, 2014-08-09 03:25:10]

On 2014/08/09 00:45:30, Dirk Pranke wrote:
> On 2014/08/09 00:12:14, eseidel wrote:
> > Note that the new infra test runner has similar support:
> > https://chromium.googlesource.com/infra/infra/+/master/test.py
> 
> Interesting, I was not aware of that tool. iannucci and I had talked about
this
> concept in the past, so maybe this is their stab at it (I had even pointed him
> at an early version of typ).
> 
> There's probably not a good reason to have two tools, so I'll take a look at
it
> in more detail  and talk to them about the overlap ...
> 
> Thanks for the pointer!

Commented on the bug which I think you filed due to this conversation. The
actual thing being run here is
https://chromium.googlesource.com/infra/testing/expect_tests, but test.py is a
wrapper for that. It's an evolved version of recipe_simulation_test.py. Yes, we
should coordinate efforts on having The Best Python Testing Framework Because
Bare unittest Sucks.

[Dirk Pranke, 2014-08-13 21:15:38]

The CQ bit was checked by dpranke@chromium.org

[commit-bot: I haz the power, 2014-08-13 21:16:06]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/dpranke@chromium.org/458643003/40001

[commit-bot: I haz the power, 2014-08-14 00:28:24]

Committed patchset #3 (40001) as 180217