Split mode incognito extension can get misleading pref changed events from regular mode

Split mode incognito extension can get misleading pref changed events from regular mode

I implemented the behavior described in comment 6 here: http://code.google.com/p/chromium/issues/detail?id=128017

BUG=128017
TEST=Manual


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=144299

[Bernhard Bauer, 2012-06-19 01:12:56]

The CL looks pretty good, but an API test would be great. This sort of stuff
tends to lead to break easily and in subtle ways, so I'd like to make sure it's
being verified automatically.

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
File chrome/browser/extensions/extension_pref_value_map.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
chrome/browser/extensions/extension_pref_value_map.cc:113: bool
ExtensionPrefValueMap::DoesExtensionControlIncognitoPref(
This method is very similar to the previous one. It might make sense to merge
them, returning the |from_incognito| value via a bool* outparam, and having
callers do the check from line 121 themselves.

You could also leave out the |incognito| parameter then, only looking at
incognito preferences if the passed in bool* is non-NULL.

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
File chrome/browser/extensions/extension_preference_helpers.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
chrome/browser/extensions/extension_preference_helpers.cc:91:
extension_service->CanCrossIncognito(*it))) {
Nit: I think this needs to be indented by one less space (it should be aligned
with the expresion above, which starts with the "!").

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
chrome/browser/extensions/extension_preference_helpers.cc:104: // b) regular
pref chanes are visible only to the incognito tabs if the
Nit: "pref changes"

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/prefs/p...
File chrome/browser/prefs/pref_notifier_impl.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/prefs/p...
chrome/browser/prefs/pref_notifier_impl.cc:7: #include "base/logging.h"
This doesn't seem necessary.

[mitchellwrosen, 2012-06-19 16:35:09]

On 2012/06/19 01:12:56, Bernhard Bauer wrote:
> The CL looks pretty good, but an API test would be great. This sort of stuff
> tends to lead to break easily and in subtle ways, so I'd like to make sure
it's
> being verified automatically.

Sure thing. I'm fairly new to writing javascript tests in Chromium... is there a
way to elegantly ensure that all pending messages are fired? Sort of like
RunMessageLoop? The reason is that I need to listen for certain events and
*fail* if they are ever received (e.g. an incognito window shouldn't receive
pref change events from a normal window if it has already set the pref itself).
Thanks!


https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/prefs/p...
> chrome/browser/prefs/pref_notifier_impl.cc:7: #include "base/logging.h"
> This doesn't seem necessary.

An artifact from me putting LOGs in everywhere, but I left this one in because I
saw other LOGs and CHECKs in the file, so base/logging.h was being transitively
included.

[Bernhard Bauer, 2012-06-19 22:09:02]

On 2012/06/19 16:35:09, mitchellwrosen wrote:
> On 2012/06/19 01:12:56, Bernhard Bauer wrote:
> > The CL looks pretty good, but an API test would be great. This sort of stuff
> > tends to lead to break easily and in subtle ways, so I'd like to make sure
> it's
> > being verified automatically.
> 
> Sure thing. I'm fairly new to writing javascript tests in Chromium... is there
a
> way to elegantly ensure that all pending messages are fired? Sort of like
> RunMessageLoop? The reason is that I need to listen for certain events and
> *fail* if they are ever received (e.g. an incognito window shouldn't receive
> pref change events from a normal window if it has already set the pref
itself).
> Thanks!

Hm. There's no built-in way to be notified when no events will fire anymore
(it's completely asynchronous and happens in the browser), but you could
probably call a Javascript method that verifies that no unwanted event was fired
from the test, after you have dispatched all events.

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/prefs/p...
> > chrome/browser/prefs/pref_notifier_impl.cc:7: #include "base/logging.h"
> > This doesn't seem necessary.
> 
> An artifact from me putting LOGs in everywhere, but I left this one in because
I
> saw other LOGs and CHECKs in the file, so base/logging.h was being
transitively
> included.

Fair enough. I guess include-what-you-use-wise it makes sense.

[mitchellwrosen, 2012-06-20 18:28:26]

Is this the kind of test you were looking for? It looks race-conditioney to me.
In fact, I'm almost positive that the listeners for bad pref change events are
removed via done() before the events are even fired. They're just a single line
of javascript apart.

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
File chrome/browser/extensions/extension_pref_value_map.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
chrome/browser/extensions/extension_pref_value_map.cc:113: bool
ExtensionPrefValueMap::DoesExtensionControlIncognitoPref(
On 2012/06/19 01:12:56, Bernhard Bauer wrote:
> This method is very similar to the previous one. It might make sense to merge
> them, returning the |from_incognito| value via a bool* outparam, and having
> callers do the check from line 121 themselves.
> 
> You could also leave out the |incognito| parameter then, only looking at
> incognito preferences if the passed in bool* is non-NULL.

Done.

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
File chrome/browser/extensions/extension_preference_helpers.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
chrome/browser/extensions/extension_preference_helpers.cc:91:
extension_service->CanCrossIncognito(*it))) {
On 2012/06/19 01:12:56, Bernhard Bauer wrote:
> Nit: I think this needs to be indented by one less space (it should be aligned
> with the expresion above, which starts with the "!").

Done.

https://chromiumcodereview.appspot.com/10559052/diff/1/chrome/browser/extensi...
chrome/browser/extensions/extension_preference_helpers.cc:104: // b) regular
pref chanes are visible only to the incognito tabs if the
On 2012/06/19 01:12:56, Bernhard Bauer wrote:
> Nit: "pref changes"

Done.

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/browser/exte...
File chrome/browser/extensions/extension_pref_value_map_unittest.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/browser/exte...
chrome/browser/extensions/extension_pref_value_map_unittest.cc:36: static const
ExtensionPrefsScope kRegular =
These "const static" -> "static const" were made to appease cpplint.py (I didn't
write this code), it was a quick fix so I just included it in this CL. Please
let me know if I should ignore things like this in the future.

[Bernhard Bauer, 2012-06-20 18:55:36]

Thanks for the test! Of course, the interesting part of this CL is what happens
with an extension in split mode, so we should add a separate test for that as
well ;-)

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/browser/exte...
File chrome/browser/extensions/extension_pref_value_map_unittest.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/browser/exte...
chrome/browser/extensions/extension_pref_value_map_unittest.cc:36: static const
ExtensionPrefsScope kRegular =
On 2012/06/20 18:28:26, mitchellwrosen wrote:
> These "const static" -> "static const" were made to appease cpplint.py (I
didn't
> write this code), it was a quick fix so I just included it in this CL. Please
> let me know if I should ignore things like this in the future.

On the contrary, thanks for the cleanup!

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/test/data/ex...
File chrome/test/data/extensions/api_test/preference/onchange/manifest.json
(right):

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/test/data/ex...
chrome/test/data/extensions/api_test/preference/onchange/manifest.json:8:
"scripts": ["test.js"]
Thanks for the cleanup here!

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/test/data/ex...
File chrome/test/data/extensions/api_test/preference/onchange/test.js (right):

https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/test/data/ex...
chrome/test/data/extensions/api_test/preference/onchange/test.js:61:
pw.thirdPartyCookiesAllowed.set({
The problem is that pw.thirdPartyCookiesAllowed.set() is an asynchronous method,
so you can't call done() immediately afterwards. Instead, you can pass it in as
the callback instead of chrome.test.callbackPass(). I think that should take
care of the race condition (as we control all the timing now).

[mitchellwrosen, 2012-06-20 19:47:02]

On 2012/06/20 18:55:36, Bernhard Bauer wrote:
https://chromiumcodereview.appspot.com/10559052/diff/8001/chrome/test/data/ex...
> chrome/test/data/extensions/api_test/preference/onchange/test.js:61:
> pw.thirdPartyCookiesAllowed.set({
> The problem is that pw.thirdPartyCookiesAllowed.set() is an asynchronous
method,
> so you can't call done() immediately afterwards. Instead, you can pass it in
as
> the callback instead of chrome.test.callbackPass(). I think that should take
> care of the race condition (as we control all the timing now).

Of course! Sometimes it's hard to tell what's boiler plate and what isn't. I
didn't even notice the chrome.test.callbackPass's

[mitchellwrosen, 2012-06-20 19:47:35]

(patch 3)

[Bernhard Bauer, 2012-06-20 20:55:27]

https://chromiumcodereview.appspot.com/10559052/diff/11001/chrome/test/data/e...
File chrome/test/data/extensions/api_test/preference/onchange/test.js (right):

https://chromiumcodereview.appspot.com/10559052/diff/11001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange/test.js:29:
chrome.test.fail("Unexpected event: " + JSON.stringify(value));
Nit: Use single quotes (I realize we did this wrong above :-D )

https://chromiumcodereview.appspot.com/10559052/diff/11001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange/test.js:35: function
openIncognito() {
What is this for?

[mitchellwrosen, 2012-06-20 21:11:41]

On 2012/06/20 20:55:27, Bernhard Bauer wrote:
>
https://chromiumcodereview.appspot.com/10559052/diff/11001/chrome/test/data/e...
https://chromiumcodereview.appspot.com/10559052/diff/11001/chrome/test/data/e...
> chrome/test/data/extensions/api_test/preference/onchange/test.js:35: function
> openIncognito() {
> What is this for?

Doesn't it run the test in split mode? I thought RunExtensionTestIncognito()
runs the test in a normal window, with the extension enabled in incognito mode,
so I just opened a new window.

[Bernhard Bauer, 2012-06-20 21:36:43]

On 2012/06/20 21:11:41, mitchellwrosen wrote:
> On 2012/06/20 20:55:27, Bernhard Bauer wrote:
> >
>
https://chromiumcodereview.appspot.com/10559052/diff/11001/chrome/test/data/e...
>
https://chromiumcodereview.appspot.com/10559052/diff/11001/chrome/test/data/e...
> > chrome/test/data/extensions/api_test/preference/onchange/test.js:35:
function
> > openIncognito() {
> > What is this for?
> 
> Doesn't it run the test in split mode? I thought RunExtensionTestIncognito()
> runs the test in a normal window, with the extension enabled in incognito
mode,
> so I just opened a new window.

Right, the test runs the extension with incognito enabled, but it's not in split
mode (the default is spanning, where the extension lives both in regular and
split mode). You're probably going to need to run an extension with a different
manifest :-/

[mitchellwrosen, 2012-06-20 21:37:33]

Could you explain the difference between split mode and spanning mode?

[Bernhard Bauer, 2012-06-20 21:46:05]

On 2012/06/20 21:37:33, mitchellwrosen wrote:
> Could you explain the difference between split mode and spanning mode?

Sure. Spanning mode means the background page is loaded once, and it receives
events for (and is able to access) both regular and incognito tabs (provided
that it is enabled for incognito mode). Split mode means that as soon as an
incognito profile is created, the background page is loaded a second time for
the incognito profile. That second background page *only* receives events for
the incognito profile, and likewise the regular background page only receives
events for the regular profile.
http://code.google.com/chrome/extensions/manifest.html#incognito also explains
it.

[mitchellwrosen, 2012-06-21 18:02:56]

Thanks. I think I need to do something like ExtensionApiTest.IncognitoSplitMode,
which runs the same extension test for both the normal and incognito processes.
The problem, though, is I need to precisely time the preference changes, (e.g.
set a regular pref both before and after the same incognito pref has been set),
instead of just running both processes asynchronously to completion.

[mitchellwrosen, 2012-06-26 01:44:04]

The test is currently passing, but printing out a couple of these:

[800:800:0625/183856:30702759031:INFO:CONSOLE(40)] "Uncaught completed", source:
apitest (40)
[800:800:0625/183856:30702759192:INFO:CONSOLE(40)] "Uncaught completed", source:
apitest (40)

I couldn't track down what they mean in the last 5 minutes (my shift is over),
but I would like to push this preliminary CL to show you the basic structure of
the test I wrote. It's a little convoluted, and please let me know if I should
change anything.

[Bernhard Bauer, 2012-06-26 08:44:29]

That's a nice test, thanks! Just mostly nits below, and a file seems to be
missing ;-)

On 2012/06/26 01:44:04, mitchellwrosen wrote:
> The test is currently passing, but printing out a couple of these:
> 
> [800:800:0625/183856:30702759031:INFO:CONSOLE(40)] "Uncaught completed",
source:
> apitest (40)
> [800:800:0625/183856:30702759192:INFO:CONSOLE(40)] "Uncaught completed",
source:
> apitest (40)

That's usually harmless. The testing code throws an exception "completed" when
the test has finished, to immediately stop running (if the test finishes inside
some callback).

> I couldn't track down what they mean in the last 5 minutes (my shift is over),
> but I would like to push this preliminary CL to show you the basic structure
of
> the test I wrote. It's a little convoluted, and please let me know if I should
> change anything.

It looks pretty good. All this synchronization is tricky, so some convolution is
expected ;-) You might want to factor out some common patterns into methods, but
that's up to you.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
File chrome/browser/extensions/extension_preference_apitest.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:7: #include
"chrome/browser/extensions/extension_test_message_listener.h"
I think you forgot to add this file to the CL?

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:181:
ASSERT_TRUE(LoadExtensionIncognito(test_data_dir_
Nit: This is a bit awkwardly broken. Could you extract the extension directory
to a local variable?

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:208:
listener_incognito5.Reply("done");
Why are you replying with "done" here instead of "ok"?

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
File chrome/test/data/extensions/api_test/preference/onchange_split/test.js
(right):

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:6: // Run
with browser_tests --gtest_filter=ExtensionApiTest.PreferenceOnChangeSplit
I think you may need to break this line.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:55:
expected[1] = {
I find expected.push() a bit nicer, as it doesn't have to rely on the fact that
there's already exactly one element in the array.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:65:
"changeDefault " + contextToString(inIncognitoContext) + " ready",
You could extract this to a variable to make the code a bit shorter.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:83:
sendMessage("changeIncognitoOnly regular listening", pass(function() {
You could directly pass in done to pass().

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:125: "
ready",
Nit: for string concatenation we usually align the continued line with the
previous one instead of indenting it four more spaces.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:139: var
done;
You could move |done| into the if-clause.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:217: if
(!inIncognitoContext) {
Nit: no braces for one-line statements.

[mitchellwrosen, 2012-06-26 17:32:52]

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
File chrome/browser/extensions/extension_preference_apitest.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:7: #include
"chrome/browser/extensions/extension_test_message_listener.h"
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> I think you forgot to add this file to the CL?
I didn't write it!

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:181:
ASSERT_TRUE(LoadExtensionIncognito(test_data_dir_
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> Nit: This is a bit awkwardly broken. Could you extract the extension directory
> to a local variable?
Done.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:208:
listener_incognito5.Reply("done");
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> Why are you replying with "done" here instead of "ok"?
Ehh, the words don't actually mean anything, since the JS never actually looks
at the message in the callback. I chose "ok" to mean, "ok, I see you're ready",
and "done" to mean, "you're done listening". They could even all be null
strings.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
File chrome/test/data/extensions/api_test/preference/onchange_split/test.js
(right):

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:6: // Run
with browser_tests --gtest_filter=ExtensionApiTest.PreferenceOnChangeSplit
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> I think you may need to break this line.
I've seen this line not broken up elsewhere (see
chromium/src/chrome/test/data/extensions/api_test/proxy/individual/test.js)

But anyways, done!

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:55:
expected[1] = {
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> I find expected.push() a bit nicer, as it doesn't have to rely on the fact
that
> there's already exactly one element in the array.

Done.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:65:
"changeDefault " + contextToString(inIncognitoContext) + " ready",
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> You could extract this to a variable to make the code a bit shorter.

Done.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:83:
sendMessage("changeIncognitoOnly regular listening", pass(function() {
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> You could directly pass in done to pass().
Do you mean like this?

sendMessage("changeIncognitoOnly regular listening", pass(done());

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:125: "
ready",
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> Nit: for string concatenation we usually align the continued line with the
> previous one instead of indenting it four more spaces.

Done.

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:217: if
(!inIncognitoContext) {
On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> Nit: no braces for one-line statements.

Done.

[Bernhard Bauer, 2012-06-26 20:09:55]

LGTM with two comments addressed. Thanks again! :)

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
File chrome/browser/extensions/extension_preference_apitest.cc (right):

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:7: #include
"chrome/browser/extensions/extension_test_message_listener.h"
On 2012/06/26 17:32:52, mitchellwrosen wrote:
> On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> > I think you forgot to add this file to the CL?
> I didn't write it!

Oh, sorry, I thought it was new 0:-)

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/browser/ext...
chrome/browser/extensions/extension_preference_apitest.cc:208:
listener_incognito5.Reply("done");
On 2012/06/26 17:32:52, mitchellwrosen wrote:
> On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> > Why are you replying with "done" here instead of "ok"?
> Ehh, the words don't actually mean anything, since the JS never actually looks
> at the message in the callback. I chose "ok" to mean, "ok, I see you're
ready",
> and "done" to mean, "you're done listening". They could even all be null
> strings.

Would you mind still changing it to "ok" (or change all strings), just so that
future readers of the code don't assume it does something special? Thanks!

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
File chrome/test/data/extensions/api_test/preference/onchange_split/test.js
(right):

https://chromiumcodereview.appspot.com/10559052/diff/17001/chrome/test/data/e...
chrome/test/data/extensions/api_test/preference/onchange_split/test.js:83:
sendMessage("changeIncognitoOnly regular listening", pass(function() {
On 2012/06/26 17:32:52, mitchellwrosen wrote:
> On 2012/06/26 08:44:29, Bernhard Bauer wrote:
> > You could directly pass in done to pass().
> Do you mean like this?
> 
> sendMessage("changeIncognitoOnly regular listening", pass(done());

just `done` -- you don't want to call |done| immediately. This is different from
pass(), because the latter tells the testing framework to keep waiting and
returns a suitable callback.

[commit-bot: I haz the power, 2012-06-26 21:13:06]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10559052/3...

[commit-bot: I haz the power, 2012-06-26 22:23:34]

Change committed as 144299