Get rid of browser_prefs::RegisterAllPrefs()

Get rid of browser_prefs::RegisterAllPrefs()

Local state and user prefs have always been separate namespaces. We'd
like to have proxy configuration in both, so let's stick to the
namespace separation and clean up testing code that incorrectly merged
the namespaces.

BUG=none
TEST=compiles and passes tests

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=75536

[Mattias Nissler (ping if slow), 2011-02-18 16:18:36]

Please review. Pawel, is the ScopedTestingBrowserProcess stuff I did for
PrefsControllerTest correct?

[danno, 2011-02-18 16:42:28]

lgtm

[Paweł Hajdan Jr., 2011-02-18 17:03:43]

http://codereview.chromium.org/6542013/diff/7001/chrome/browser/ui/cocoa/opti...
File chrome/browser/ui/cocoa/options/preferences_window_controller_unittest.mm
(right):

http://codereview.chromium.org/6542013/diff/7001/chrome/browser/ui/cocoa/opti...
chrome/browser/ui/cocoa/options/preferences_window_controller_unittest.mm:45:
browser_process_.get()->SetPrefService(&local_state_);
Just use g_browser_process here.

http://codereview.chromium.org/6542013/diff/7001/chrome/test/testing_browser_...
File chrome/test/testing_browser_process.h (right):

http://codereview.chromium.org/6542013/diff/7001/chrome/test/testing_browser_...
chrome/test/testing_browser_process.h:154: TestingBrowserProcess* get() {
That seems wrong, I think the code should just use g_browser_process.

[Mattias Nissler (ping if slow), 2011-02-18 17:08:04]

Pawel, I put comments for your consideration. Thanks!

http://codereview.chromium.org/6542013/diff/7001/chrome/browser/ui/cocoa/opti...
File chrome/browser/ui/cocoa/options/preferences_window_controller_unittest.mm
(right):

http://codereview.chromium.org/6542013/diff/7001/chrome/browser/ui/cocoa/opti...
chrome/browser/ui/cocoa/options/preferences_window_controller_unittest.mm:45:
browser_process_.get()->SetPrefService(&local_state_);
On 2011/02/18 17:03:43, Paweł Hajdan Jr. wrote:
> Just use g_browser_process here.

That won't work, since the SetPrefService is a special helper for testing. It
doesn't exist on BrowserProcess, but only TestingBrowserProcess. So working with
g_browser_process I'd have to cast, which I think is good to avoid. Also, what's
the problem with having a scoped browser process and being able to configure it
for testing?

http://codereview.chromium.org/6542013/diff/7001/chrome/test/testing_browser_...
File chrome/test/testing_browser_process.h (right):

http://codereview.chromium.org/6542013/diff/7001/chrome/test/testing_browser_...
chrome/test/testing_browser_process.h:154: TestingBrowserProcess* get() {
On 2011/02/18 17:03:43, Paweł Hajdan Jr. wrote:
> That seems wrong, I think the code should just use g_browser_process.

See my other comment for why this is here. Would you rather like the typecast?

[Paweł Hajdan Jr., 2011-02-18 19:39:29]

A-ha! LGTM.

I remember a solution in other unit test which uses the cast, but after thinking
about it for a while the solution from this patch sounds better. In fact, I'll
take that into account in my TestingBrowserProcess-related cleanup.