Fix up GCAPI's LaunchGoogleChrome() method.

Fix up GCAPI's LaunchGoogleChrome() method to use the current way of finding the path to chrome.exe.

Fix LaunchGoogleChromeWithDimensions such that it looks for the correctly named window for current Chromes.

Add an in_background parameter to LaunchGoogleChromeWithDimensions() that will place the launched Chrome window behind all other windows.

Remove usage of ATL (was pulled in only for CComPtr).

BUG=NONE
TEST=NONE
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=113016

[grt (UTC plus 2), 2011-12-05 15:14:22]

LGTM with nits

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi.cc
File chrome/installer/gcapi/gcapi.cc (right):

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:41: using base::Time;
what do you think about moving these above RegKey so that the sorting is akin to
the #include order?

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:446: if
(SUCCEEDED(ipl.CreateInstance(__uuidof(ProcessLauncherClass),
what, !FAILED implies SUCCEEDED to you?

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:474: handle = FindWindowEx(NULL, handle,
L"Chrome_WidgetWin_0", NULL);
how about making a kChromeWindowClass for this?

[Roger Tawa OOO till Jul 10th, 2011-12-05 16:19:45]

lgtm

One question and nit below.  Nice cleanup of the code!

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi.cc
File chrome/installer/gcapi/gcapi.cc (right):

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:357: if (!IsChromeInstalled(install_key)) {
For my own info, why try HKLM before HKCU?

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:383: }
You don't want to do another path exists check here, instead of relying on the
code below to eventually fail?

[robertshield, 2011-12-05 18:18:00]

Thanks, ptal

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi.cc
File chrome/installer/gcapi/gcapi.cc (right):

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:41: using base::Time;
On 2011/12/05 15:14:22, grt wrote:
> what do you think about moving these above RegKey so that the sorting is akin
to
> the #include order?

I think that's a splendid idea.

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:357: if (!IsChromeInstalled(install_key)) {
On 2011/12/05 16:19:45, Roger Tawa wrote:
> For my own info, why try HKLM before HKCU?

If both exist and the user-level one is launched, it will self-destruct and
delegate to the system-level one. I figured this process might interact badly
with the LaunchGoogleChromeWithDimensions code below.

I could also just never launch the user-level one (like the old code), but Mike
stated that this code could also be used with user-level Chrome installations so
I made this change.

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:383: }
On 2011/12/05 16:19:45, Roger Tawa wrote:
> You don't want to do another path exists check here, instead of relying on the
> code below to eventually fail?

Good idea, done.

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:446: if
(SUCCEEDED(ipl.CreateInstance(__uuidof(ProcessLauncherClass),
On 2011/12/05 15:14:22, grt wrote:
> what, !FAILED implies SUCCEEDED to you?

#define SUCCEEDED(hr) (((HRESULT)(hr)) >= 0)
#define FAILED(hr) (((HRESULT)(hr)) < 0)

yes. 

I suppose there could be degrees of failure, DIDNT_REALLY_WORK(), FAILED(),
BLEW_UP_APOCALYPTICALLY(), etc.. but I'm sticking with the glass-half-full
outlook on life.

http://codereview.chromium.org/8776038/diff/3002/chrome/installer/gcapi/gcapi...
chrome/installer/gcapi/gcapi.cc:474: handle = FindWindowEx(NULL, handle,
L"Chrome_WidgetWin_0", NULL);
On 2011/12/05 15:14:22, grt wrote:
> how about making a kChromeWindowClass for this?

Done.

[grt (UTC plus 2), 2011-12-05 19:37:11]

still LGTM

[Roger Tawa OOO till Jul 10th, 2011-12-05 20:02:40]

lgtm