Fix regression where component extensions were being loaded for login profile

Fix regression where component extensions were being loaded for login profile


BUG=chromium-os:18953
TEST=Login on production build and check for extra extension processes

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=96597

[Denis Lagno, 2011-08-12 06:50:02]

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
File chrome/browser/profiles/profile_manager.cc (right):

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:315: bool go_off_the_record = true;
if you want to restore the behaviour that existed before my erroneous CL then
you need to make it false here.

[Dmitry Polukhin, 2011-08-12 08:04:31]

Dave, thank you for fixing this regression.

LGTM, feel free to commit after changing initialization to false (it looks like
it is just misprint, now go_off_the_record is true all the time in your code).

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
File chrome/browser/profiles/profile_manager.cc (right):

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:207:
command_line.HasSwitch(switches::kLoginProfile)) {
There are 3 places with identical no trivial check and ifdef. I think it makes
sense to move it into separate function with comment as above and TODO.

#1

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:208: // Don't init extensions for
this profile
This comment has no sense here anymore. It should be in that function with
check.

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:320:
command_line.HasSwitch(switches::kLoginProfile))) {
#2

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:399:
command_line.HasSwitch(switches::kLoginProfile))) {
#3

[zel, 2011-08-12 16:15:35]

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
File chrome/browser/profiles/profile_manager.cc (right):

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:321: go_off_the_record = true;
go_off_the_record is always true!?

[DaveMoore, 2011-08-12 16:37:42]

xiyuan, could you test this w/ the webui path? I'm not sure how to determine if
it breaks something.

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
File chrome/browser/profiles/profile_manager.cc (right):

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:207:
command_line.HasSwitch(switches::kLoginProfile)) {
I combined two into a call, but one had some logic within the check so I left
it.
On 2011/08/12 08:04:31, Dmitry Polukhin wrote:
> There are 3 places with identical no trivial check and ifdef. I think it makes
> sense to move it into separate function with comment as above and TODO.
> 
> #1

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:208: // Don't init extensions for
this profile
On 2011/08/12 08:04:31, Dmitry Polukhin wrote:
> This comment has no sense here anymore. It should be in that function with
> check.

Done.

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:315: bool go_off_the_record = true;
On 2011/08/12 06:50:02, Denis Lagno wrote:
> if you want to restore the behaviour that existed before my erroneous CL then
> you need to make it false here.

Done.

http://codereview.chromium.org/7630002/diff/1/chrome/browser/profiles/profile...
chrome/browser/profiles/profile_manager.cc:321: go_off_the_record = true;
On 2011/08/12 16:15:40, zel wrote:
> go_off_the_record is always true!?

Done.

[xiyuan, 2011-08-12 17:36:48]

LGTM

Tried the CL and see gaia_auth extension still working.