Temporary workaround to get tests that reference singletons either directly o...

Temporary workaround to get tests that reference singletons either directly or indirectly to pass.
With the recent Singleton refactoring, these tests started exposing some inherent problems with singletons, namely that one test can reference a state set by a different test, which can cause all sorts of flakiness.
Furthermore, some of these tests could reference the registry which may have some user specific settings that could cause the tests to fail

To work around these issues, I'm adding a few utility classes for tests (registry virtualization) and exposing a method to reset singletons before a test is run.

TEST=Fixes flakiness and a few tests that could fail on the waterfall.
BUG=none


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=68701

[amit, 2010-12-08 23:40:34]

lgtm, thanks for fixing the tests!

http://codereview.chromium.org/5564009/diff/1/chrome_frame/test/urlmon_monike...
File chrome_frame/test/urlmon_moniker_unittest.cc (right):

http://codereview.chromium.org/5564009/diff/1/chrome_frame/test/urlmon_monike...
chrome_frame/test/urlmon_moniker_unittest.cc:25: DeleteAllSingletons();
It safe now to delete the moniker patch code along with the tests :)

[tommi (sloooow) - chröme, 2010-12-08 23:47:36]

thanks.  Do you mind if I remove the moniker patching code on a separate
change? (since it involves editing gyp files too)

On Wed, Dec 8, 2010 at 6:40 PM, <amit@chromium.org> wrote:

> lgtm, thanks for fixing the tests!
>
>
>
>
http://codereview.chromium.org/5564009/diff/1/chrome_frame/test/urlmon_monike...
> File chrome_frame/test/urlmon_moniker_unittest.cc (right):
>
>
>
http://codereview.chromium.org/5564009/diff/1/chrome_frame/test/urlmon_monike...
> chrome_frame/test/urlmon_moniker_unittest.cc:25: DeleteAllSingletons();
> It safe now to delete the moniker patch code along with the tests :)
>
>
> http://codereview.chromium.org/5564009/
>

[amit, 2010-12-09 00:11:22]

yeah separate CL is the way to go. If we are on it then I think we can get
rid of OnHttpEquiv and related goo as well.

On Wed, Dec 8, 2010 at 3:47 PM, Tommi <tommi@chromium.org> wrote:

> thanks.  Do you mind if I remove the moniker patching code on a separate
> change? (since it involves editing gyp files too)
>
>
> On Wed, Dec 8, 2010 at 6:40 PM, <amit@chromium.org> wrote:
>
>> lgtm, thanks for fixing the tests!
>>
>>
>>
>>
http://codereview.chromium.org/5564009/diff/1/chrome_frame/test/urlmon_monike...
>> File chrome_frame/test/urlmon_moniker_unittest.cc (right):
>>
>>
>>
http://codereview.chromium.org/5564009/diff/1/chrome_frame/test/urlmon_monike...
>> chrome_frame/test/urlmon_moniker_unittest.cc:25: DeleteAllSingletons();
>> It safe now to delete the moniker patch code along with the tests :)
>>
>>
>> http://codereview.chromium.org/5564009/
>>
>
>