Remove inappropriate chrome.{extension,app} bindings for platform apps

Remove inappropriate chrome.{extension,app} bindings for platform apps.

Also:

-Begin process of moving from chrome.extension.lastError to chrome.runtime.lastError.

-Add chrome.runtime.getManifest and chrome.runtime.getURL



BUG=131717
TEST=In platform apps, chrome.extension and chrome.app should not contain any
extension or legacy hosted app stuff on them.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=145238

[asargent_no_longer_on_chrome, 2012-06-25 18:45:00]

mihaip: please review whole CL

aa, kalman: FYI, drive-bys welcome

[not at google - send to devlin, 2012-06-25 23:35:56]

General wishful thinking comment: if we're deprecating
chrome.extension.lastError then ideally we could actually replace it with a more
sensible system, e.g. having the error as part of every callback (there is a
design doc somewhere).

I presume that this i, emergency surgery for platform apps or something, though.
Sad.

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
File chrome/common/extensions/api/extension_api.cc (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:637: // apps pick up features we
don't want via dependencies.
Can we do this in ChromeV8Context::GetAvailableExtensionAPIs? I'd rather keep
this kind of specific information out of this file, and besides, it might
regress in here (e.g. if we convert to features system).

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:734: if
(extension->is_platform_app() &&
ditto

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
File chrome/common/extensions/api/runtime.json (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:18: },
description?

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:30: "type": "any"
Aren't these always strings?

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:68: "optional": true,
Why optional?

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
File chrome/renderer/resources/extensions/runtime_custom_bindings.js (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
chrome/renderer/resources/extensions/runtime_custom_bindings.js:26: return
appNatives.GetDetails();
Perhaps cache this result?

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
File chrome/renderer/resources/extensions/send_request.js (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
chrome/renderer/resources/extensions/send_request.js:22: };
chromeHidden is not the place for these methods. Either put them on exports, or
even better, not as part of send_request.js at all and in their own module. E.g.
last_error.js. (hopefully we can avoid the chrome.runtime = chrome.runtime ||
{}, why is that necessary anyway?)

http://codereview.chromium.org/10666027/diff/8001/chrome/test/data/extensions...
File chrome/test/data/extensions/platform_apps/restrictions/main.js (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/test/data/extensions...
chrome/test/data/extensions/platform_apps/restrictions/main.js:165: return;
It might be useful to have separate checks for each of these, it would make it
easier to track down regressions.

[Mihai Parparita -not on Chrome, 2012-06-25 23:45:18]

LGTM

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
File chrome/common/extensions/api/extension_api.cc (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:639: feature_name == "extension")
Can this also check for feature_name == "app", like the other check, in case
something ends up with a dependency on apps?

Better yet, add a IsFeatureAllowedForExtension(feature_name, extension) helper
and use it in both?

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:733: // Platform apps don't want
the extension or legacy app API bindings.
Can there be a TODO about this being specifiable more declaratively via the
feature system?

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
File chrome/common/extensions/api/runtime.json (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:7: "namespace": "runtime",
Should this also include lastError?

[not at google - send to devlin, 2012-06-25 23:55:11]

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
File chrome/renderer/resources/extensions/runtime_custom_bindings.js (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
chrome/renderer/resources/extensions/runtime_custom_bindings.js:26: return
appNatives.GetDetails();
On 2012/06/25 23:35:56, kalman wrote:
> Perhaps cache this result?

Eh, never mind caching it.

But I just noticed that you're using appNatives here not a new runtimeNatives
class. The problem with appNatives is that it also puts an "id" thing in the
manifest. I had a separate change quite similar to this one which put the id
separately on chrome.runtime.id and I prefer that to including in getManifest().

The other problem is that... well, it's appNatives not runtimeNatives. That's
confusing.

http://codereview.chromium.org/10640013/

Copy runtime_natives.cc from there?

[Aaron Boodman, 2012-06-26 01:27:41]

On Mon, Jun 25, 2012 at 4:35 PM,  <[email protected]> wrote:
> General wishful thinking comment: if we're deprecating
> chrome.extension.lastError then ideally we could actually replace it with a
> more
> sensible system, e.g. having the error as part of every callback (there is a
> design doc somewhere).

I don't see any deprecation of anything here - just aliasing
chrome.extension.lastError at chrome.runtime.lastError, right?

That said, I think I've said this before, but I think that all new
APIs should pass the error as the first param to the callback.

Mihai, what do you think about making this a launch blocker for the
beta/stable?

- a

[Mihai Parparita -not on Chrome, 2012-06-26 05:35:15]

On Mon, Jun 25, 2012 at 6:27 PM, Aaron Boodman <[email protected]> wrote:

> That said, I think I've said this before, but I think that all new
> APIs should pass the error as the first param to the callback.
>
> Mihai, what do you think about making this a launch blocker for the
> beta/stable?
>

Sure. Though I'm not convinced that having it as the first param is that
readable (or clean, given legacy APIs). What about the other approaches
considered in
https://docs.google.com/a/google.com/document/d/1ZJtY80XK0eU9Xs1dBD2ZqoeJK9FP...
param, keep lastError but make its presence/type signaled via the
schema, so that the docs include it)?

Mihai

[Aaron Boodman, 2012-06-26 05:55:00]

On Mon, Jun 25, 2012 at 10:35 PM, Mihai Parparita <[email protected]> wrote:
> On Mon, Jun 25, 2012 at 6:27 PM, Aaron Boodman <[email protected]> wrote:
>>
>> That said, I think I've said this before, but I think that all new
>> APIs should pass the error as the first param to the callback.
>>
>> Mihai, what do you think about making this a launch blocker for the
>> beta/stable?>
>
> Sure. Though I'm not convinced that having it as the first param is that
> readable (or clean, given legacy APIs). What about the other approaches
> considered in
>
https://docs.google.com/a/google.com/document/d/1ZJtY80XK0eU9Xs1dBD2ZqoeJK9FP...
> (last param, keep lastError but make its presence/type signaled via the
> schema, so that the docs include it)?

Sorry, to be more clear, my current favorite proposal is:

- keep lastError
- make it first-class via the schema system, and make it show up in docs
- new APIs also pass it as first param
- in new APIs, make lastError a richer object for the cases where that
would be useful

The problem with making it the last param is that it slowly becomes
less consistent as the APIs age.

wdyt?

- a

[not at google - send to devlin, 2012-06-26 22:40:30]

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
File chrome/common/extensions/api/extension_api.cc (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:602:
ResolveDependencies(&temp_result);
if (extension->is_platform_app())
  RemoveNonPlatformAppAPIs(&temp_result);

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:637: // apps pick up features we
don't want via dependencies.
On 2012/06/25 23:35:56, kalman wrote:
> Can we do this in ChromeV8Context::GetAvailableExtensionAPIs? I'd rather keep
> this kind of specific information out of this file, and besides, it might
> regress in here (e.g. if we convert to features system).

I changed my mind here, a better way to avoid regressions would be to do
something like what we do for removing privileged APIs in the
UNBLESSED/CONTENT_SCRIPT cases. I've added a comment where. This would replace
the code here and below.

[asargent_no_longer_on_chrome, 2012-06-28 23:10:33]

Addressed comments - PTAL.

Aaron/Mihai - I didn't address the bigger picture question of whether we want to
switch new APIs to have error as the first argument and skip having
chrome.runtime.lastError for them. Perhaps we can hash out whether we agree on
that as the plan and we can do that as a follow up CL in the next week or two.

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
File chrome/common/extensions/api/extension_api.cc (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:637: // apps pick up features we
don't want via dependencies.
On 2012/06/26 22:40:30, kalman wrote:
> On 2012/06/25 23:35:56, kalman wrote:
> > Can we do this in ChromeV8Context::GetAvailableExtensionAPIs? I'd rather
keep
> > this kind of specific information out of this file, and besides, it might
> > regress in here (e.g. if we convert to features system).
> 
> I changed my mind here, a better way to avoid regressions would be to do
> something like what we do for removing privileged APIs in the
> UNBLESSED/CONTENT_SCRIPT cases. I've added a comment where. This would replace
> the code here and below.

Ok, I've done it this way.

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:639: feature_name == "extension")
On 2012/06/25 23:45:18, Mihai Parparita wrote:
> Can this also check for feature_name == "app", like the other check, in case
> something ends up with a dependency on apps?
> 
> Better yet, add a IsFeatureAllowedForExtension(feature_name, extension) helper
> and use it in both?

Done. (up above though)

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/extension_api.cc:733: // Platform apps don't want
the extension or legacy app API bindings.
On 2012/06/25 23:45:18, Mihai Parparita wrote:
> Can there be a TODO about this being specifiable more declaratively via the
> feature system?
Removed this code, but added the TODO in other place above.

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
File chrome/common/extensions/api/runtime.json (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:7: "namespace": "runtime",
On 2012/06/25 23:45:18, Mihai Parparita wrote:
> Should this also include lastError?

Yes. Done.

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:18: },
On 2012/06/25 23:35:56, kalman wrote:
> description?

Done.

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:30: "type": "any"
On 2012/06/25 23:35:56, kalman wrote:
> Aren't these always strings?

Yes, good catch. Fixed.

http://codereview.chromium.org/10666027/diff/8001/chrome/common/extensions/ap...
chrome/common/extensions/api/runtime.json:68: "optional": true,
On 2012/06/25 23:35:56, kalman wrote:
> Why optional?

Good catch. I copied this from app.json, where I guess it can return null if
called from not within an app. But for chrome.runtime it should always return a
value.

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
File chrome/renderer/resources/extensions/runtime_custom_bindings.js (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
chrome/renderer/resources/extensions/runtime_custom_bindings.js:26: return
appNatives.GetDetails();
On 2012/06/25 23:55:11, kalman wrote:
> On 2012/06/25 23:35:56, kalman wrote:
> > Perhaps cache this result?
> 
> Eh, never mind caching it.
> 
> But I just noticed that you're using appNatives here not a new runtimeNatives
> class. The problem with appNatives is that it also puts an "id" thing in the
> manifest. I had a separate change quite similar to this one which put the id
> separately on chrome.runtime.id and I prefer that to including in
getManifest().
> 
> The other problem is that... well, it's appNatives not runtimeNatives. That's
> confusing.
> 
> http://codereview.chromium.org/10640013/
> 
> Copy runtime_natives.cc from there?

Done.

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
File chrome/renderer/resources/extensions/send_request.js (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/renderer/resources/e...
chrome/renderer/resources/extensions/send_request.js:22: };
On 2012/06/25 23:35:56, kalman wrote:
> chromeHidden is not the place for these methods. Either put them on exports,
or
> even better, not as part of send_request.js at all and in their own module.
E.g.
> last_error.js. (hopefully we can avoid the chrome.runtime = chrome.runtime ||
> {}, why is that necessary anyway?)

Done (Moved these to last_error.js)

http://codereview.chromium.org/10666027/diff/8001/chrome/test/data/extensions...
File chrome/test/data/extensions/platform_apps/restrictions/main.js (right):

http://codereview.chromium.org/10666027/diff/8001/chrome/test/data/extensions...
chrome/test/data/extensions/platform_apps/restrictions/main.js:165: return;
On 2012/06/25 23:35:56, kalman wrote:
> It might be useful to have separate checks for each of these, it would make it
> easier to track down regressions.
Refactored to use assertEq for each one.

[not at google - send to devlin, 2012-06-29 02:10:28]

lgtm with comments.

http://codereview.chromium.org/10666027/diff/13004/chrome/common/extensions/a...
File chrome/common/extensions/api/extension_api.cc (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/common/extensions/a...
chrome/common/extensions/api/extension_api.cc:591: if (extension &&
!IsFeatureAllowedForExtension(*i, *extension)) {
This method should never get passed a NULL extension, so CHECK rather than guard
against it?

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
File chrome/renderer/extensions/extension_dispatcher.cc (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
chrome/renderer/extensions/extension_dispatcher.cc:40: #include
"chrome/renderer/extensions/runtime_custom_bindings.h"
nit: r after p

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
chrome/renderer/extensions/extension_dispatcher.cc:710: if (!extension ||
!extension->is_platform_app())
Something is going wrong if there isn't an extension here (right? or am I going
crazy?). So CHECK?

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
File chrome/renderer/extensions/runtime_custom_bindings.cc (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
chrome/renderer/extensions/runtime_custom_bindings.cc:32: 
Again, something going wrong if there isn't an extension here (my bad for
putting that in the other patch). CHECK?

http://codereview.chromium.org/10666027/diff/13004/chrome/test/data/extension...
File chrome/test/data/extensions/platform_apps/runtime/test.js (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/test/data/extension...
chrome/test/data/extensions/platform_apps/runtime/test.js:9:
chrome.test.runTests([
This isn't really a platform-app specific test, it belongs more in
c/t/d/e/api_test?

http://codereview.chromium.org/10666027/diff/13004/chrome/test/data/extension...
chrome/test/data/extensions/platform_apps/runtime/test.js:36:
assertEq(manifest.app.background.scripts[0], "test.js");
Just assertEq(manifest.app.background.scripts, ["test.js"])

[asargent_no_longer_on_chrome, 2012-07-02 17:29:43]

Comments addressed, hitting CQ button now.

http://codereview.chromium.org/10666027/diff/13004/chrome/common/extensions/a...
File chrome/common/extensions/api/extension_api.cc (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/common/extensions/a...
chrome/common/extensions/api/extension_api.cc:591: if (extension &&
!IsFeatureAllowedForExtension(*i, *extension)) {
On 2012/06/29 02:10:28, kalman wrote:
> This method should never get passed a NULL extension, so CHECK rather than
guard
> against it?

Good point. Changed from Extension* to Extension&.

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
File chrome/renderer/extensions/extension_dispatcher.cc (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
chrome/renderer/extensions/extension_dispatcher.cc:40: #include
"chrome/renderer/extensions/runtime_custom_bindings.h"
On 2012/06/29 02:10:28, kalman wrote:
> nit: r after p

Done.

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
chrome/renderer/extensions/extension_dispatcher.cc:710: if (!extension ||
!extension->is_platform_app())
On 2012/06/29 02:10:28, kalman wrote:
> Something is going wrong if there isn't an extension here (right? or am I
going
> crazy?). So CHECK?

It's not immediately obvious that the Extension* is guaranteed to be non-null
from a quick look at the surrounding code and the implementation of
ClassifyJavaScriptContext. I'll put in the check and we'll see if it catches
anything, but perhaps a larger refactoring is in order to split this into the "I
have an extension" (where we can use const Extension&) and "I don't have an
extension" cases.

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
File chrome/renderer/extensions/runtime_custom_bindings.cc (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/renderer/extensions...
chrome/renderer/extensions/runtime_custom_bindings.cc:32: 
On 2012/06/29 02:10:28, kalman wrote:
> Again, something going wrong if there isn't an extension here (my bad for
> putting that in the other patch). CHECK?

Done.

http://codereview.chromium.org/10666027/diff/13004/chrome/test/data/extension...
File chrome/test/data/extensions/platform_apps/runtime/test.js (right):

http://codereview.chromium.org/10666027/diff/13004/chrome/test/data/extension...
chrome/test/data/extensions/platform_apps/runtime/test.js:9:
chrome.test.runTests([
On 2012/06/29 02:10:28, kalman wrote:
> This isn't really a platform-app specific test, it belongs more in
> c/t/d/e/api_test?

Done.

http://codereview.chromium.org/10666027/diff/13004/chrome/test/data/extension...
chrome/test/data/extensions/platform_apps/runtime/test.js:36:
assertEq(manifest.app.background.scripts[0], "test.js");
On 2012/06/29 02:10:28, kalman wrote:
> Just assertEq(manifest.app.background.scripts, ["test.js"])

Done.

[commit-bot: I haz the power, 2012-07-02 17:30:05]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10666027/15010

[commit-bot: I haz the power, 2012-07-02 18:25:17]

Try job failure for 10666027-15010 (retry) on linux_rel for step
"interactive_ui_tests".
It's a second try, previously, step "interactive_ui_tests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_rel&...

[commit-bot: I haz the power, 2012-07-02 18:28:51]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10666027/15010

[commit-bot: I haz the power, 2012-07-02 19:09:36]

Try job failure for 10666027-15010 (retry) on linux_rel for step
"interactive_ui_tests".
It's a second try, previously, step "interactive_ui_tests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_rel&...

[commit-bot: I haz the power, 2012-07-02 23:25:43]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10666027/15010

[commit-bot: I haz the power, 2012-07-03 01:32:26]

Change committed as 145238