Add an option to the test_shell to enable it to load each test several times

We have sees bugs which could have been caught by reloading a page one or more times and finding that the results for one of the subsequent loads is not the same as the expected output. The new option--multiple-loads=X will load each test X time. The output dumped by test_shell is the result of the last load. The default when the option is specified without a value is 2 times in debug mode and 5 times in release mode.

To be able to have more fine-grained control of how the JavaScript engine behaves for each load the existing flag --js-flags can now specify a list of flag-sets like this

  --js-flags="--xxx,--noxxx --yyy,--noyyy"

First load will run with --xxx, the second with --yyy and the third without any (the 'no' prefix is handled by V8 to turn off the flag).

The changes to the Python test runner will be in a separate change.

BUG=none
TEST=none
Review URL: http://codereview.chromium.org/4016002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@63863 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 6 insertions, 4 deletions

diff --git a/webkit/tools/test_shell/test_shell_mac.mm b/webkit/tools/test_shell/test_shell_mac.mm
index d2b90a7..30eab0f 100644
--- a/webkit/tools/test_shell/test_shell_mac.mm
+++ b/webkit/tools/test_shell/test_shell_mac.mm
@@ -370,10 +370,12 @@ void TestShell::TestFinished() {
     return;  // reached when running under test_shell_tests
 
   test_is_pending_ = false;
-  NSWindow* window = *(TestShell::windowList()->begin());
-  WindowMap::iterator it = window_map_.Get().find(window);
-  if (it != window_map_.Get().end())
-    TestShell::Dump(it->second);
+  if (dump_when_finished_) {
+    NSWindow* window = *(TestShell::windowList()->begin());
+    WindowMap::iterator it = window_map_.Get().find(window);
+    if (it != window_map_.Get().end())
+      TestShell::Dump(it->second);
+  }
   MessageLoop::current()->Quit();
 }