New installation validator machinery to check the machine state....

New installation validator machinery to check the machine state.  This CL contains the first two parts of the validator:

1. installer::InstallationValidator class, used by setup.exe to confirm that all is well following an installer operation.  Violations are logged at ERROR level.
2. validate_installation.exe, used by humans to check the overall state of a machine.

The third part, in which violations fire nonfatal test failures for use in other tests, will follow in a subsequent CL.

BUG=none
TEST=Run validate_installation.exe on machine(s) containing all conceivable permutations of Chrome and Chrome Frame, and see if it reports any violations.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=74819

[tommi (sloooow) - chröme, 2011-02-12 00:11:17]

lgtm!  This cl is all sorts of excellent!

http://codereview.chromium.org/6490024/diff/5001/chrome/chrome_installer.gypi
File chrome/chrome_installer.gypi (right):

http://codereview.chromium.org/6490024/diff/5001/chrome/chrome_installer.gypi...
chrome/chrome_installer.gypi:72: '<(DEPTH)/testing/gmock.gyp:gmock',
I believe this should be above the gtest line

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/tools/valid...
File chrome/installer/tools/validate_installation_main.cc (right):

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/tools/valid...
chrome/installer/tools/validate_installation_main.cc:9: #include <cstdio>
I have a sneaky suspicion that chromium has an iostream preference.  I'm not
sure it matters here though

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
File chrome/installer/util/installation_validator.cc (right):

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
chrome/installer/util/installation_validator.cc:351: bool rock_on = true;
\mm/

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
File chrome/installer/util/installation_validator.h (right):

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
chrome/installer/util/installation_validator.h:79: typedef
std::vector<std::pair<std::string, bool>> SwitchExpectations;
space before the last >

[amit, 2011-02-12 01:08:19]

drive by bow to the awesomeness :)
Thanks!

[grt (UTC plus 2), 2011-02-14 04:05:20]

Thanks for the comments, guys.

http://codereview.chromium.org/6490024/diff/5001/chrome/chrome_installer.gypi
File chrome/chrome_installer.gypi (right):

http://codereview.chromium.org/6490024/diff/5001/chrome/chrome_installer.gypi...
chrome/chrome_installer.gypi:72: '<(DEPTH)/testing/gmock.gyp:gmock',
On 2011/02/12 00:11:17, tommi wrote:
> I believe this should be above the gtest line

Done.

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/tools/valid...
File chrome/installer/tools/validate_installation_main.cc (right):

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/tools/valid...
chrome/installer/tools/validate_installation_main.cc:9: #include <cstdio>
On 2011/02/12 00:11:17, tommi wrote:
> I have a sneaky suspicion that chromium has an iostream preference.  I'm not
> sure it matters here though

I used iostreams initially, but either cpplint or the presubmit scripts didn't
like them as much as I did.  The Google C++ guide says not to use them beyond
logging.  I don't see mention of them in the Chromium style guide, so I suppose
fprintf is the right thing here.

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
File chrome/installer/util/installation_validator.h (right):

http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
chrome/installer/util/installation_validator.h:79: typedef
std::vector<std::pair<std::string, bool>> SwitchExpectations;
On 2011/02/12 00:11:17, tommi wrote:
> space before the last >

Done.

[tommi (sloooow) - chröme, 2011-02-14 05:15:33]

lgtm.

On iostream, logging is what I was thinking of, so fprintf is cool.

On Sun, Feb 13, 2011 at 11:05 PM, <grt@chromium.org> wrote:

> Thanks for the comments, guys.
>
>
>
>
> http://codereview.chromium.org/6490024/diff/5001/chrome/chrome_installer.gypi
> File chrome/chrome_installer.gypi (right):
>
>
>
http://codereview.chromium.org/6490024/diff/5001/chrome/chrome_installer.gypi...
> chrome/chrome_installer.gypi:72: '<(DEPTH)/testing/gmock.gyp:gmock',
> On 2011/02/12 00:11:17, tommi wrote:
>
>> I believe this should be above the gtest line
>>
>
> Done.
>
>
>
>
http://codereview.chromium.org/6490024/diff/5001/chrome/installer/tools/valid...
> File chrome/installer/tools/validate_installation_main.cc (right):
>
>
>
http://codereview.chromium.org/6490024/diff/5001/chrome/installer/tools/valid...
> chrome/installer/tools/validate_installation_main.cc:9: #include
> <cstdio>
> On 2011/02/12 00:11:17, tommi wrote:
>
>> I have a sneaky suspicion that chromium has an iostream preference.
>>
> I'm not
>
>> sure it matters here though
>>
>
> I used iostreams initially, but either cpplint or the presubmit scripts
> didn't like them as much as I did.  The Google C++ guide says not to use
> them beyond logging.  I don't see mention of them in the Chromium style
> guide, so I suppose fprintf is the right thing here.
>
>
>
>
http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
> File chrome/installer/util/installation_validator.h (right):
>
>
>
http://codereview.chromium.org/6490024/diff/5001/chrome/installer/util/instal...
> chrome/installer/util/installation_validator.h:79: typedef
> std::vector<std::pair<std::string, bool>> SwitchExpectations;
> On 2011/02/12 00:11:17, tommi wrote:
>
>> space before the last >
>>
>
> Done.
>
>
> http://codereview.chromium.org/6490024/
>

[robertshield, 2011-02-14 14:09:11]

LGTM, nice work!

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/installer_...
File chrome/installer/installer_tools.gyp (right):

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/installer_...
chrome/installer/installer_tools.gyp:3: 'version_py':
'../../chrome/tools/build/version.py',
I vaguely remember gyp having a pre-defined variable called DEPTH that could be
used to make this kind of thing stable across folder moves. Something like
'version_py': '<(DEPTH>/chrome/tools/build/version.py',
http://code.google.com/p/gyp/wiki/InputFormatReference

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/tools/vali...
File chrome/installer/tools/validate_installation_main.cc (right):

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/tools/vali...
chrome/installer/tools/validate_installation_main.cc:63:
LOG_ASSERT(old_message_handler_ == NULL);
does LOG_ASSERT DCHECK?

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/util/insta...
File chrome/installer/util/installation_validator.cc (right):

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/util/insta...
chrome/installer/util/installation_validator.cc:59:
product_state.uninstall_command().HasSwitch(switches::kMultiInstall);
is_multi_install appears unused?

[robertshield, 2011-02-14 14:09:13]

LGTM, nice work!

[grt (UTC plus 2), 2011-02-14 15:48:03]

Thanks.  I've uploaded the final rev that I'm committing.

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/installer_...
File chrome/installer/installer_tools.gyp (right):

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/installer_...
chrome/installer/installer_tools.gyp:3: 'version_py':
'../../chrome/tools/build/version.py',
On 2011/02/14 14:09:11, robertshield wrote:
> I vaguely remember gyp having a pre-defined variable called DEPTH that could
be
> used to make this kind of thing stable across folder moves. Something like
> 'version_py': '<(DEPTH>/chrome/tools/build/version.py',
> http://code.google.com/p/gyp/wiki/InputFormatReference

Cool.  Done.

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/tools/vali...
File chrome/installer/tools/validate_installation_main.cc (right):

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/tools/vali...
chrome/installer/tools/validate_installation_main.cc:63:
LOG_ASSERT(old_message_handler_ == NULL);
On 2011/02/14 14:09:11, robertshield wrote:
> does LOG_ASSERT DCHECK?

#define LOG_ASSERT(condition)  \
  LOG_IF(FATAL, !(condition)) << "Assert failed: " #condition ". "

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/util/insta...
File chrome/installer/util/installation_validator.cc (right):

http://codereview.chromium.org/6490024/diff/11001/chrome/installer/util/insta...
chrome/installer/util/installation_validator.cc:59:
product_state.uninstall_command().HasSwitch(switches::kMultiInstall);
On 2011/02/14 14:09:11, robertshield wrote:
> is_multi_install appears unused?

Nice catch.  Thanks.