Proxy settings automation hooks and related proxy settings fixes.

Adds two Python/PyAuto functions, GetProxySettings() and SetProxySetting(), that allow tests to read and manipulate Chrome OS proxy settings. 

Fixes issue with proxy settings where they reset when changing only the port of a proxy without changing the URL. This happens when using the automation hooks because they create a new ProxyCrosSettingsProvider with each invocation. This is done by removing the caching behavior of ProxyCrosSettingsProvider.

Modifies the type of the "Port" textboxes in the Proxy Settings UI to Integer so that ProxyCrosSettingsProvider uses a consistent type for both Get() and Set().


BUG=chromium-os:13650, chromium-os:12727
TEST=Manual. Test included for GetProxySettings() that will run in the Autotest lab. stanleyw will add additional tests that use SetProxySetting().

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=80537

[dtu, 2011-04-04 21:03:36]

+nirnimesh for Python and C++
+xiyuan for C++
+dhg for HTML and JS

PTAL

[dtu, 2011-04-04 21:10:34]

Apparently dhg isn't around anymore. xiyuan, can you also take a look at the
HTML/JS?

[xiyuan, 2011-04-04 21:10:57]

On 2011/04/04 21:10:34, Dave Tu wrote:
> Apparently dhg isn't around anymore. xiyuan, can you also take a look at the
> HTML/JS?

Sure.

[xiyuan, 2011-04-04 21:24:20]

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
File chrome/browser/chromeos/proxy_cros_settings_provider.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:61: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
You need AppendPortIfValid to get cached user input because underlying proxy
service does not remember invalid input.
e.g.
if user enters port first without a valid uri, then port will be lost and revert
to default port.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:67: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
similar, you need FormServerUriIfValid.

We might want to get it a better name though. FormServerUriIfValid and
AppendPortIfValid is not intutive and does not tell you what they do.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/resources/op...
File chrome/browser/resources/options/chromeos_proxy.html (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/resources/op...
chrome/browser/resources/options/chromeos_proxy.html:35: <input
id="proxyHostSinglePort" type="text" dataType="number" size="5"
pref="cros.session.proxy.singlehttpport" disabled>
I don't think we need a new "dataType" attribute. If you need an integer, you
could change type="text" to type="number". That should bind to PrefNumber and
give you an integer back. And webkit would also validate user's input as well.

[Nirnimesh, 2011-04-04 21:38:27]

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
File chrome/browser/automation/testing_automation_provider.h (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
chrome/browser/automation/testing_automation_provider.h:1039: void
SetProxySetting(DictionaryValue* args, IPC::Message* reply_message);
append 's' to the method name

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
chrome/browser/automation/testing_automation_provider_chromeos.cc:48: delete
setting_dict;
who allocs setting_dict?

http://codereview.chromium.org/6737035/diff/10011/chrome/test/functional/chro...
File chrome/test/functional/chromeos_wifi.py (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/test/functional/chro...
chrome/test/functional/chromeos_wifi.py:25: def testProxySettings(self):
testGetProxySettings

http://codereview.chromium.org/6737035/diff/10011/chrome/test/pyautolib/pyaut...
File chrome/test/pyautolib/pyauto.py (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/test/pyautolib/pyaut...
chrome/test/pyautolib/pyauto.py:2546: 'type' : int - Type of proxy. Should be
one of:
remove the space before :
Repeat below

[stanleyw, 2011-04-04 22:13:27]

http://codereview.chromium.org/6737035/diff/10011/chrome/test/pyautolib/pyaut...
File chrome/test/pyautolib/pyauto.py (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/test/pyautolib/pyaut...
chrome/test/pyautolib/pyauto.py:2522: u'single': False,
For error throwing, type as a string instead of int is easier to decode.

[dtu, 2011-04-04 22:48:56]

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
File chrome/browser/automation/testing_automation_provider.h (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
chrome/browser/automation/testing_automation_provider.h:1039: void
SetProxySetting(DictionaryValue* args, IPC::Message* reply_message);
On 2011/04/04 21:38:27, Nirnimesh wrote:
> append 's' to the method name

Done.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
chrome/browser/automation/testing_automation_provider_chromeos.cc:48: delete
setting_dict;
On 2011/04/04 21:38:27, Nirnimesh wrote:
> who allocs setting_dict?

Good catch, this is a mistake. setting_dict is still owned by the settings
provider.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
File chrome/browser/chromeos/proxy_cros_settings_provider.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:61: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
On 2011/04/04 21:24:20, xiyuan wrote:
> You need AppendPortIfValid to get cached user input because underlying proxy
> service does not remember invalid input.
> e.g.
> if user enters port first without a valid uri, then port will be lost and
revert
> to default port.

This was true previously, when this function was creating a URI string and using
FromURI to parse a ProxyServer from that string. But now it uses the ProxyServer
constructor directly and will happily accept arbitrary URIs.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:67: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
On 2011/04/04 21:24:20, xiyuan wrote:
> similar, you need FormServerUriIfValid.
> 
> We might want to get it a better name though. FormServerUriIfValid and
> AppendPortIfValid is not intutive and does not tell you what they do.

See above.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/resources/op...
File chrome/browser/resources/options/chromeos_proxy.html (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/resources/op...
chrome/browser/resources/options/chromeos_proxy.html:35: <input
id="proxyHostSinglePort" type="text" dataType="number" size="5"
pref="cros.session.proxy.singlehttpport" disabled>
On 2011/04/04 21:24:20, xiyuan wrote:
> I don't think we need a new "dataType" attribute. If you need an integer, you
> could change type="text" to type="number". That should bind to PrefNumber and
> give you an integer back. And webkit would also validate user's input as well.

I tried type="number". Having the input validation is nice, and you can set min
and max values; however, using this element changes the UI in two ways that are
not preferable:
1. It adds up/down arrows into the box, which are not useful when entering a
port number.
2. When you take focus off the box, it automatically adds thousands separators,
e.g. 12345 becomes 12,345.
#2 is a bigger problem. I thought there might be some way to change the behavior
using CSS or JS but couldn't find any documentation on it since it is a new
HTML5 element and this seems to happen only in recent versions of Chrome.

http://codereview.chromium.org/6737035/diff/10011/chrome/test/functional/chro...
File chrome/test/functional/chromeos_wifi.py (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/test/functional/chro...
chrome/test/functional/chromeos_wifi.py:25: def testProxySettings(self):
On 2011/04/04 21:38:27, Nirnimesh wrote:
> testGetProxySettings

Done.

http://codereview.chromium.org/6737035/diff/10011/chrome/test/pyautolib/pyaut...
File chrome/test/pyautolib/pyauto.py (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/test/pyautolib/pyaut...
chrome/test/pyautolib/pyauto.py:2522: u'single': False,
On 2011/04/04 22:13:27, stanleyw wrote:
> For error throwing, type as a string instead of int is easier to decode.

Added GetProxyTypeName(). Hope that is sufficient.

http://codereview.chromium.org/6737035/diff/10011/chrome/test/pyautolib/pyaut...
chrome/test/pyautolib/pyauto.py:2546: 'type' : int - Type of proxy. Should be
one of:
On 2011/04/04 21:38:27, Nirnimesh wrote:
> remove the space before :
> Repeat below

Done.

[stanleyw, 2011-04-04 23:30:47]

.py code LGTM

[xiyuan, 2011-04-04 23:31:18]

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
chrome/browser/automation/testing_automation_provider_chromeos.cc:48: delete
setting_dict;
On 2011/04/04 22:48:57, Dave Tu wrote:
> On 2011/04/04 21:38:27, Nirnimesh wrote:
> > who allocs setting_dict?
> 
> Good catch, this is a mistake. setting_dict is still owned by the settings
> provider.

emm... Acutally, the comment in cros_settings_provider.h for Get is wrong. The
caller owns the returned value. 

Could you keep this logic and update the comments there?

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
File chrome/browser/chromeos/proxy_cros_settings_provider.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:61: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
On 2011/04/04 22:48:57, Dave Tu wrote:
> On 2011/04/04 21:24:20, xiyuan wrote:
> > You need AppendPortIfValid to get cached user input because underlying proxy
> > service does not remember invalid input.
> > e.g.
> > if user enters port first without a valid uri, then port will be lost and
> revert
> > to default port.
> 
> This was true previously, when this function was creating a URI string and
using
> FromURI to parse a ProxyServer from that string. But now it uses the
ProxyServer
> constructor directly and will happily accept arbitrary URIs.

It would be great that the backend proxy service would remeber arbitrary values.

Please make sure you test this case:
1. Go to proxy settings and enter an invalid host name (simulate user typo);
2. Enter a valid port number;
3. Click anywhere on settings page to trigger a blur event so that the change is
saved;
4. Refresh proxy settings page and see if the port number is correct.

If that works, then I am okay.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/resources/op...
File chrome/browser/resources/options/chromeos_proxy.html (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/resources/op...
chrome/browser/resources/options/chromeos_proxy.html:35: <input
id="proxyHostSinglePort" type="text" dataType="number" size="5"
pref="cros.session.proxy.singlehttpport" disabled>
On 2011/04/04 22:48:57, Dave Tu wrote:
> On 2011/04/04 21:24:20, xiyuan wrote:
> > I don't think we need a new "dataType" attribute. If you need an integer,
you
> > could change type="text" to type="number". That should bind to PrefNumber
and
> > give you an integer back. And webkit would also validate user's input as
well.
> 
> I tried type="number". Having the input validation is nice, and you can set
min
> and max values; however, using this element changes the UI in two ways that
are
> not preferable:
> 1. It adds up/down arrows into the box, which are not useful when entering a
> port number.
> 2. When you take focus off the box, it automatically adds thousands
separators,
> e.g. 12345 becomes 12,345.
> #2 is a bigger problem. I thought there might be some way to change the
behavior
> using CSS or JS but couldn't find any documentation on it since it is a new
> HTML5 element and this seems to happen only in recent versions of Chrome.

Okay. In this case, I am okay with the new attribute approach.

[dtu, 2011-04-05 00:04:11]

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/automation/t...
chrome/browser/automation/testing_automation_provider_chromeos.cc:48: delete
setting_dict;
On 2011/04/04 23:31:18, xiyuan wrote:
> On 2011/04/04 22:48:57, Dave Tu wrote:
> > On 2011/04/04 21:38:27, Nirnimesh wrote:
> > > who allocs setting_dict?
> > 
> > Good catch, this is a mistake. setting_dict is still owned by the settings
> > provider.
> 
> emm... Acutally, the comment in cros_settings_provider.h for Get is wrong. The
> caller owns the returned value. 
> 
> Could you keep this logic and update the comments there?

Done. Thanks for the heads up.

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
File chrome/browser/chromeos/proxy_cros_settings_provider.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:61: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
On 2011/04/04 23:31:18, xiyuan wrote:
> On 2011/04/04 22:48:57, Dave Tu wrote:
> > On 2011/04/04 21:24:20, xiyuan wrote:
> > > You need AppendPortIfValid to get cached user input because underlying
proxy
> > > service does not remember invalid input.
> > > e.g.
> > > if user enters port first without a valid uri, then port will be lost and
> > revert
> > > to default port.
> > 
> > This was true previously, when this function was creating a URI string and
> using
> > FromURI to parse a ProxyServer from that string. But now it uses the
> ProxyServer
> > constructor directly and will happily accept arbitrary URIs.
> 
> It would be great that the backend proxy service would remeber arbitrary
values.
> 
> Please make sure you test this case:
> 1. Go to proxy settings and enter an invalid host name (simulate user typo);
> 2. Enter a valid port number;
> 3. Click anywhere on settings page to trigger a blur event so that the change
is
> saved;
> 4. Refresh proxy settings page and see if the port number is correct.
> 
> If that works, then I am okay.

I have tried it and it works =)

[xiyuan, 2011-04-05 16:32:49]

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
File chrome/browser/chromeos/proxy_cros_settings_provider.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:61: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
On 2011/04/05 00:04:11, Dave Tu wrote:
> On 2011/04/04 23:31:18, xiyuan wrote:
> > On 2011/04/04 22:48:57, Dave Tu wrote:
> > > On 2011/04/04 21:24:20, xiyuan wrote:
> > > > You need AppendPortIfValid to get cached user input because underlying
> proxy
> > > > service does not remember invalid input.
> > > > e.g.
> > > > if user enters port first without a valid uri, then port will be lost
and
> > > revert
> > > > to default port.
> > > 
> > > This was true previously, when this function was creating a URI string and
> > using
> > > FromURI to parse a ProxyServer from that string. But now it uses the
> > ProxyServer
> > > constructor directly and will happily accept arbitrary URIs.
> > 
> > It would be great that the backend proxy service would remeber arbitrary
> values.
> > 
> > Please make sure you test this case:
> > 1. Go to proxy settings and enter an invalid host name (simulate user typo);
> > 2. Enter a valid port number;
> > 3. Click anywhere on settings page to trigger a blur event so that the
change
> is
> > saved;
> > 4. Refresh proxy settings page and see if the port number is correct.
> > 
> > If that works, then I am okay.
> 
> I have tried it and it works =)

Cool. Thanks for trying this out.

Then we don't need a local cache for the user typed values at all. Please remove
cache_ and its references. Or at least put a TODO for this.

[dtu, 2011-04-05 18:48:48]

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
File chrome/browser/chromeos/proxy_cros_settings_provider.cc (right):

http://codereview.chromium.org/6737035/diff/10011/chrome/browser/chromeos/pro...
chrome/browser/chromeos/proxy_cros_settings_provider.cc:61: val,
config.single_proxy, net::ProxyServer::SCHEME_HTTP));
On 2011/04/05 16:32:49, xiyuan wrote:
> On 2011/04/05 00:04:11, Dave Tu wrote:
> > On 2011/04/04 23:31:18, xiyuan wrote:
> > > On 2011/04/04 22:48:57, Dave Tu wrote:
> > > > On 2011/04/04 21:24:20, xiyuan wrote:
> > > > > You need AppendPortIfValid to get cached user input because underlying
> > proxy
> > > > > service does not remember invalid input.
> > > > > e.g.
> > > > > if user enters port first without a valid uri, then port will be lost
> and
> > > > revert
> > > > > to default port.
> > > > 
> > > > This was true previously, when this function was creating a URI string
and
> > > using
> > > > FromURI to parse a ProxyServer from that string. But now it uses the
> > > ProxyServer
> > > > constructor directly and will happily accept arbitrary URIs.
> > > 
> > > It would be great that the backend proxy service would remeber arbitrary
> > values.
> > > 
> > > Please make sure you test this case:
> > > 1. Go to proxy settings and enter an invalid host name (simulate user
typo);
> > > 2. Enter a valid port number;
> > > 3. Click anywhere on settings page to trigger a blur event so that the
> change
> > is
> > > saved;
> > > 4. Refresh proxy settings page and see if the port number is correct.
> > > 
> > > If that works, then I am okay.
> > 
> > I have tried it and it works =)
> 
> Cool. Thanks for trying this out.
> 
> Then we don't need a local cache for the user typed values at all. Please
remove
> cache_ and its references. Or at least put a TODO for this.

Huh, I thought I had already done that. Done now.

[xiyuan, 2011-04-05 18:50:01]

LGTM

Thanks for making all the changes. :)

[Nirnimesh, 2011-04-05 19:27:53]

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:48: delete
setting_dict;
delete setting, not setting dict, since that's the object that was returned by
settings_provider.Get().

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:314:
!args->Get("value", &value)) {
this might fit on previous line

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:322:
chromeos::ProxyCrosSettingsProvider().Set(setting_path, value->DeepCopy());
who deallocs the DeepCopy()?

http://codereview.chromium.org/6737035/diff/7007/chrome/test/pyautolib/pyauto.py
File chrome/test/pyautolib/pyauto.py (right):

http://codereview.chromium.org/6737035/diff/7007/chrome/test/pyautolib/pyauto...
chrome/test/pyautolib/pyauto.py:2517: return values.get(proxy_type, 'Unknown
proxy type')
Remove the default arg. It should return None in case of invalid proxy_type.
Better still, return values[proxy_type] so that this crashes in case of invalid
proxy_type

http://codereview.chromium.org/6737035/diff/7007/chrome/test/pyautolib/pyauto...
chrome/test/pyautolib/pyauto.py:2551: Owner must be logged in for these to take
effect.
Did you mean 'take effect' or persist?
So if a non-owner user logs in, he won't be able to change proxy settings?

[xiyuan, 2011-04-05 19:44:32]

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:322:
chromeos::ProxyCrosSettingsProvider().Set(setting_path, value->DeepCopy());
On 2011/04/05 19:27:53, Nirnimesh wrote:
> who deallocs the DeepCopy()?

cros settings provider will own the value passed int via Set and will release it
on its destruction.

[Nirnimesh, 2011-04-05 19:49:13]

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:322:
chromeos::ProxyCrosSettingsProvider().Set(setting_path, value->DeepCopy());
On 2011/04/05 19:44:32, xiyuan wrote:
> On 2011/04/05 19:27:53, Nirnimesh wrote:
> > who deallocs the DeepCopy()?
> 
> cros settings provider will own the value passed int via Set and will release
it
> on its destruction.

Ah, I now see the comment in cros_settings_provider.h. Thanks

[dtu, 2011-04-05 21:29:10]

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
File chrome/browser/automation/testing_automation_provider_chromeos.cc (right):

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:48: delete
setting_dict;
On 2011/04/05 19:27:53, Nirnimesh wrote:
> delete setting, not setting dict, since that's the object that was returned by
> settings_provider.Get().

Done.

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:314:
!args->Get("value", &value)) {
On 2011/04/05 19:27:53, Nirnimesh wrote:
> this might fit on previous line

Done.

http://codereview.chromium.org/6737035/diff/7007/chrome/browser/automation/te...
chrome/browser/automation/testing_automation_provider_chromeos.cc:322:
chromeos::ProxyCrosSettingsProvider().Set(setting_path, value->DeepCopy());
Added a comment here, just in case.

http://codereview.chromium.org/6737035/diff/7007/chrome/test/pyautolib/pyauto.py
File chrome/test/pyautolib/pyauto.py (right):

http://codereview.chromium.org/6737035/diff/7007/chrome/test/pyautolib/pyauto...
chrome/test/pyautolib/pyauto.py:2517: return values.get(proxy_type, 'Unknown
proxy type')
On 2011/04/05 19:27:53, Nirnimesh wrote:
> Remove the default arg. It should return None in case of invalid proxy_type.
> Better still, return values[proxy_type] so that this crashes in case of
invalid
> proxy_type

Done.

http://codereview.chromium.org/6737035/diff/7007/chrome/test/pyautolib/pyauto...
chrome/test/pyautolib/pyauto.py:2551: Owner must be logged in for these to take
effect.
On 2011/04/05 19:27:53, Nirnimesh wrote:
> Did you mean 'take effect' or persist?
> So if a non-owner user logs in, he won't be able to change proxy settings?

Fixed.

[Nirnimesh, 2011-04-05 21:39:11]

LGTM, with one minor comment.

http://codereview.chromium.org/6737035/diff/9010/chrome/test/pyautolib/pyauto.py
File chrome/test/pyautolib/pyauto.py (right):

http://codereview.chromium.org/6737035/diff/9010/chrome/test/pyautolib/pyauto...
chrome/test/pyautolib/pyauto.py:2519: def GetProxySettings(self):
To reduce the confusion, please add 'OnChromeos' in these function names. These
functions won't work for the desktops right?