Do machine inspection on install....

Do machine inspection on install.

When installing Chrome [x]or Chrome Frame in multi mode, check if the other product (CF xor Chrome) is installed.
If it is, also include it in the installation as we need to update both products in sync.

For Chrome Frame, I also added checks to see if CEEE is already enabled or if the user has already opted into GCF.
If CEEE has been enabled, we implicitly include it when the install is run even though ceee is not on the command line.
If ready-mode is set on the command line but Chrome Frame is already installed without it, we won't enable ready mode.

TEST=Try installing Chrome Frame in multi mode.  Then install a more recent version of Chrome in multi mode.  Both product should be updated.  Secondly, for Chrome Frame, install it first and then try to run the installer again, but with ready-mode ont the command line.  Chrome Frame should not be switched to ready mode.
BUG=61609
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=70268

[grt (UTC plus 2), 2010-12-29 16:51:50]

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/setup/setu...
File chrome/installer/setup/setup_main.cc (right):

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/setup/setu...
chrome/installer/setup/setup_main.cc:732: VLOG(1) << "Install distribution:
Chrome";
It might be useful for the log to indicate that a product is being implicitly
added to the mapping.

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/chrom...
File chrome/installer/util/chrome_frame_distribution.cc (right):

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/chrom...
chrome/installer/util/chrome_frame_distribution.cc:45: LOG(INFO) << "CEEE is not
specified on the command line but CEEE is "
This log message says that CEEE is already installed, but isn't that only the
case if the call to uninstall.HasSwitch below returns true?

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/helpe...
File chrome/installer/util/helper.cc (right):

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/helpe...
chrome/installer/util/helper.cc:64: LOG(ERROR) << "No uninstallation arguments
for "
Are these two ERROR conditions bad enough that we should abort installation?  In
other words, will all callers do something that will be non-harmful in these
cases?

[tommi (sloooow) - chröme, 2010-12-29 17:30:09]

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/setup/setu...
File chrome/installer/setup/setup_main.cc (right):

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/setup/setu...
chrome/installer/setup/setup_main.cc:732: VLOG(1) << "Install distribution:
Chrome";
On 2010/12/29 16:51:50, grt wrote:
> It might be useful for the log to indicate that a product is being implicitly
> added to the mapping.

Added a LOG_IF(INFO, *installed) to the loop above.

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/chrom...
File chrome/installer/util/chrome_frame_distribution.cc (right):

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/chrom...
chrome/installer/util/chrome_frame_distribution.cc:45: LOG(INFO) << "CEEE is not
specified on the command line but CEEE is "
On 2010/12/29 16:51:50, grt wrote:
> This log message says that CEEE is already installed, but isn't that only the
> case if the call to uninstall.HasSwitch below returns true?

Thanks for catching this.  Must have happened when I was shuffling the code
around a bit

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/helpe...
File chrome/installer/util/helper.cc (right):

http://codereview.chromium.org/6091008/diff/20001/chrome/installer/util/helpe...
chrome/installer/util/helper.cc:64: LOG(ERROR) << "No uninstallation arguments
for "
On 2010/12/29 16:51:50, grt wrote:
> Are these two ERROR conditions bad enough that we should abort installation? 
In
> other words, will all callers do something that will be non-harmful in these
> cases?

I was thinking the same but in the end opted on the side of "things must be
pretty messed up for this to happen and if the installer is being run, it will
likely fix it".
It's also non trivial to back out from here.

[grt (UTC plus 2), 2010-12-29 18:57:21]

Looks groovy to me.