Launch Google Update on uninstall.

Launch Google Update on uninstall.

So it can self-destruct if Chrome (or Chrome Frame) was the last Omaha-managed product on the system.

BUG=114786
TEST=Install Google Update 1.3.21.105 or later and Chrome w/ --verbose-logging.  Uninstall Chrome and see that Google Update vanishes immediately.  Take a look at chrome_installer.log and look for messages related to Launching Google Update's uninstaller.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=126888

[grt (UTC plus 2), 2012-03-13 20:03:18]

Ganesh: I assume that Chrome's uninstaller should wait for Google Update to exit
before terminating.  Could Google Update take a terribly long time to uninstall
itself?  If so, should Chrome's uninstaller wait for a reasonable amount of time
and then move on?

[erikwright (departed), 2012-03-13 20:55:20]

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/set...
File chrome/installer/setup/setup_main.cc (right):

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/set...
chrome/installer/setup/setup_main.cc:850: if (base::LaunchProcess(uninstall_cmd,
launch_options, &process)) {
Can you use ScopedHandle::Receive instead of an interim base::ProcessHandle? It
returns a HANDLE*, but ProcessHandle is typedef'd to HANDLE anyway.

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/uti...
File chrome/installer/util/google_update_settings_unittest.cc (right):

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/uti...
chrome/installer/util/google_update_settings_unittest.cc:582: EXPECT_TRUE(
EXPECT_EQ? Then upon failure you would see what was actually returned.

Ditto below.

[S. Ganesh, 2012-03-14 04:16:38]

On 2012/03/13 20:03:18, grt wrote:
> Ganesh: I assume that Chrome's uninstaller should wait for Google Update to
exit
> before terminating.  Could Google Update take a terribly long time to
uninstall
> itself?  If so, should Chrome's uninstaller wait for a reasonable amount of
time
> and then move on?

Hi Greg, yes, waiting for a reasonable amount of time and moving on should be
fine. In the typical case, this should be quick (seconds). There are cases where
Omaha uninstall might wait for other processes to terminate and take longer
(around a minute).

[grt (UTC plus 2), 2012-03-14 13:47:02]

Thanks for the comments; PTAL.  Chrome's uninstaller will now wait up to 20s for
Omaha to exit.  Ganesh: does this seem like a reasonable amount of time to you? 
Is there any way for us to track how much time Omaha actually takes (is this
reported up in some kind of an uninstall ping)?

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/set...
File chrome/installer/setup/setup_main.cc (right):

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/set...
chrome/installer/setup/setup_main.cc:850: if (base::LaunchProcess(uninstall_cmd,
launch_options, &process)) {
On 2012/03/13 20:55:20, erikwright wrote:
> Can you use ScopedHandle::Receive instead of an interim base::ProcessHandle?
It
> returns a HANDLE*, but ProcessHandle is typedef'd to HANDLE anyway.

Awesome, thanks for the pointer (Receive() didn't exist the last time I used
ScopedHandle).

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/uti...
File chrome/installer/util/google_update_settings_unittest.cc (right):

https://chromiumcodereview.appspot.com/9693055/diff/2001/chrome/installer/uti...
chrome/installer/util/google_update_settings_unittest.cc:582: EXPECT_TRUE(
On 2012/03/13 20:55:20, erikwright wrote:
> EXPECT_EQ? Then upon failure you would see what was actually returned.

Excellent idea, thanks.

[S. Ganesh, 2012-03-14 22:59:37]

On 2012/03/14 13:47:02, grt wrote:
> Thanks for the comments; PTAL.  Chrome's uninstaller will now wait up to 20s
for
> Omaha to exit.  Ganesh: does this seem like a reasonable amount of time to
you? 
> Is there any way for us to track how much time Omaha actually takes (is this
> reported up in some kind of an uninstall ping)?

Hi Greg, 20 seconds looks reasonable. The time Omaha took to uninstall is not
reported at the moment.

[erikwright (departed), 2012-03-15 01:52:08]

LGTM.

[commit-bot: I haz the power, 2012-03-15 01:57:49]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/grt@chromium.org/9693055/6001

[commit-bot: I haz the power, 2012-03-15 04:16:58]

Try job failure for 9693055-6001 (retry) on win_rel for step "browser_tests".
It's a second try, previously, step "browser_tests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2012-03-15 10:40:35]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/grt@chromium.org/9693055/6001

[commit-bot: I haz the power, 2012-03-15 12:00:32]

Change committed as 126888