Implements frame sniffer to watch iframe loading state.

Implements frame sniffer to watch iframe loading state.

Also sniffer listener was implemented on the WebUI side, which signals when
GAIA page loading error appears.

BUG=chromium-os:18744
TEST=Behind captive portal, try to click 'Add new user'. Corresponding message page should be shown.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=99777

[altimofeev, 2011-08-31 14:43:39]

PTAL:
John: RenderView related part.
Xiyan: JS part.

Note, I haven't tested this with the real captive portal (device for testing is
being delivered). But I have tried it with other types of network errors: when
GAIA page can't be loaded, but connection exists.

[Nikita (slow), 2011-08-31 15:29:15]

http://codereview.chromium.org/7779010/diff/6002/chrome/app/generated_resourc...
File chrome/app/generated_resources.grd (right):

http://codereview.chromium.org/7779010/diff/6002/chrome/app/generated_resourc...
chrome/app/generated_resources.grd:13174: Before signing in, please start a
Guest session to activate the network or choose another one.
Might also use same message that is shown in offline help.
http://www.google.com/support/chromeos/bin/answer.py?hl=en&answer=1059250&ctx...

"Connected to a network, but still can't sign in? You might be on a network that
requires a web-based authentication. Start a guest browsing session (if
available) and see if a network log-in page appears. Once you're logged in to
the network and are able to visit different web pages, exit the guest session
and try signing in again."

[xiyuan, 2011-08-31 16:25:35]

I am not sure if DidFailProvisionalLoad would give a good test for captive
portal. When you are in a portal, you probably will get a activation page for
any URL you visit. Would DidFailProvisionalLoad be called in this case?

Also, flimflam detects portal case as well and put STATE_PROTAL as network
state. In this case, navigator.onLine is false (it is only true for
STATE_ONLINE).  You won't be able to get to Gaia UI at all.

http://codereview.chromium.org/7779010/diff/6002/chrome/browser/resources/chr...
File chrome/browser/resources/chromeos/login/screen_offline_message.html
(right):

http://codereview.chromium.org/7779010/diff/6002/chrome/browser/resources/chr...
chrome/browser/resources/chromeos/login/screen_offline_message.html:5: <span
id="captive-portal-message-text" hidden
i18n-content="captivePortalMessage"></span>
nit: wrap and keep the line <= 80 chars

[jam, 2011-08-31 16:36:41]

this doesn't need to be in content at all, so all the renderer side should be
moved to chrome. you can then dispatch the message to start this in
chrome_render_view_observer.cc. see
http://dev.chromium.org/developers/content-module for more information about
content/chrome split

http://codereview.chromium.org/7779010/diff/6002/content/common/view_messages.h
File content/common/view_messages.h (right):

http://codereview.chromium.org/7779010/diff/6002/content/common/view_messages...
content/common/view_messages.h:2104:
IPC_MESSAGE_ROUTED1(ViewMsg_StartFrameSniffer,
the messages are separated by ViewMsg first, then ViewHostMsg

[jam, 2011-08-31 16:58:15]

Also, for future reference, IPCs shouldn't span the content\chrome boundary. so
anytime you find code that sends an IPC from content and receives it in chrome,
or vice versa, that's a big hint that something isn't right. Ananta is working
on removing all such instances, and after that's done, we will use DEPS to
ensure this isn't possible.

On 2011/08/31 16:36:41, John Abd-El-Malek wrote:
> this doesn't need to be in content at all, so all the renderer side should be
> moved to chrome. you can then dispatch the message to start this in
> chrome_render_view_observer.cc. see
> http://dev.chromium.org/developers/content-module for more information about
> content/chrome split

[altimofeev, 2011-09-02 15:26:36]

Thank you for your comments. Please take another look.

> I am not sure if DidFailProvisionalLoad would give a good test for captive
> portal. When you are in a portal, you probably will get a activation page for
> any URL you visit. Would DidFailProvisionalLoad be called in this case?
> Also, flimflam detects portal case as well and put STATE_PROTAL as network
> state. In this case, navigator.onLine is false (it is only true for
> STATE_ONLINE).  You won't be able to get to Gaia UI at all.

Yes, you are right. There is captive portal detection in network library. But,
AFAIK it checks google.com, so it is useless in case, when "google.com:80" is
opened. On the other hand, this CL makes Chrome suggest to login as Guest first
instead of showing 'request timeout error page' in the "gaia-frame" iframe.

The main idea behind this CL is to prevent "error page" to be shown instead of
GAIA web page. I'm planning to improve captive portal notification with future
CLs (i.e. separate offline and captive portal cases on login screen).

http://codereview.chromium.org/7779010/diff/6002/chrome/app/generated_resourc...
File chrome/app/generated_resources.grd (right):

http://codereview.chromium.org/7779010/diff/6002/chrome/app/generated_resourc...
chrome/app/generated_resources.grd:13174: Before signing in, please start a
Guest session to activate the network or choose another one.
On 2011/08/31 15:29:15, Nikita Kostylev wrote:
> Might also use same message that is shown in offline help.
>
http://www.google.com/support/chromeos/bin/answer.py?hl=en&answer=1059250&ctx...
> 
> "Connected to a network, but still can't sign in? You might be on a network
that
> requires a web-based authentication. Start a guest browsing session (if
> available) and see if a network log-in page appears. Once you're logged in to
> the network and are able to visit different web pages, exit the guest session
> and try signing in again."

Done.

http://codereview.chromium.org/7779010/diff/6002/chrome/browser/resources/chr...
File chrome/browser/resources/chromeos/login/screen_offline_message.html
(right):

http://codereview.chromium.org/7779010/diff/6002/chrome/browser/resources/chr...
chrome/browser/resources/chromeos/login/screen_offline_message.html:5: <span
id="captive-portal-message-text" hidden
i18n-content="captivePortalMessage"></span>
On 2011/08/31 16:25:36, xiyuan wrote:
> nit: wrap and keep the line <= 80 chars

Done.

http://codereview.chromium.org/7779010/diff/6002/content/common/view_messages.h
File content/common/view_messages.h (right):

http://codereview.chromium.org/7779010/diff/6002/content/common/view_messages...
content/common/view_messages.h:2104:
IPC_MESSAGE_ROUTED1(ViewMsg_StartFrameSniffer,
On 2011/08/31 16:36:41, John Abd-El-Malek wrote:
> the messages are separated by ViewMsg first, then ViewHostMsg

Done.

[xiyuan, 2011-09-02 15:49:53]

LGTM the chromeos part.

http://codereview.chromium.org/7779010/diff/15001/chrome/browser/resources/ch...
File chrome/browser/resources/chromeos/login/screen_offline_message.js (right):

http://codereview.chromium.org/7779010/diff/15001/chrome/browser/resources/ch...
chrome/browser/resources/chromeos/login/screen_offline_message.js:96: * @param
{integer} error Error code.
nit: integer -> number

[jam, 2011-09-02 16:52:59]

http://codereview.chromium.org/7779010/diff/6002/content/common/view_messages.h
File content/common/view_messages.h (right):

http://codereview.chromium.org/7779010/diff/6002/content/common/view_messages...
content/common/view_messages.h:2104:
IPC_MESSAGE_ROUTED1(ViewMsg_StartFrameSniffer,
On 2011/09/02 15:26:36, altimofeev wrote:
> On 2011/08/31 16:36:41, John Abd-El-Malek wrote:
> > the messages are separated by ViewMsg first, then ViewHostMsg
> 
> Done.

sorry maybe I should have been more verbose :) What I mean is you'll notice that
all the ViewMsg messages come first, then all the ViewHostMsg. so if you add two
messages that are different, they won't be beside each other. i.e.
ViewMsg_StartFrameSniffer would go in line 1250.

but, since this is a message that's sent and received in chrome, it doesn't
belong in content. so it should be moved to chrome\common\render_messages.h and
be called ChromeViewHostMsg and ChromeViewMsg like the other messages there
(that's what I meant when I said that content shouldn't be changed at all)

[altimofeev, 2011-09-05 09:39:33]

Zel: ping
John: thank you for the explanations. Done.
Xiyuan: thanks. Done.

http://codereview.chromium.org/7779010/diff/15001/chrome/browser/resources/ch...
File chrome/browser/resources/chromeos/login/screen_offline_message.js (right):

http://codereview.chromium.org/7779010/diff/15001/chrome/browser/resources/ch...
chrome/browser/resources/chromeos/login/screen_offline_message.js:96: * @param
{integer} error Error code.
On 2011/09/02 15:49:53, xiyuan wrote:
> nit: integer -> number

Done.

[zel, 2011-09-05 17:15:43]

LGTM

http://codereview.chromium.org/7779010/diff/23001/chrome/renderer/chrome_rend...
File chrome/renderer/chrome_render_view_observer.cc (right):

http://codereview.chromium.org/7779010/diff/23001/chrome/renderer/chrome_rend...
chrome/renderer/chrome_render_view_observer.cc:242:
IPC_MESSAGE_HANDLER(ChromeViewMsg_StartFrameSniffer, OnStartFrameSniffer)
#if defined(OS_CHROMEOS)?

[altimofeev, 2011-09-06 14:47:52]

http://codereview.chromium.org/7779010/diff/23001/chrome/renderer/chrome_rend...
File chrome/renderer/chrome_render_view_observer.cc (right):

http://codereview.chromium.org/7779010/diff/23001/chrome/renderer/chrome_rend...
chrome/renderer/chrome_render_view_observer.cc:242:
IPC_MESSAGE_HANDLER(ChromeViewMsg_StartFrameSniffer, OnStartFrameSniffer)
On 2011/09/05 17:15:44, zel wrote:
> #if defined(OS_CHROMEOS)?

Done.

[jam, 2011-09-06 15:41:53]

lgtm