Add KeepAliveStateObserver, add the Restart option

Add KeepAliveStateObserver, that allows clients to get
notified of state changes in the KeepAliveRegistry.
The first supported option is KeepAliveRestartOption.

This CL does not add any observer to act on that yet.

BUG=587926, 585080
Committed: https://crrev.com/3563ecaf39e0c08836a17743d5233de1e0e7b6f0
Cr-Commit-Position: refs/heads/master@{#380174}

[dgn, 2016-02-23 16:58:45]

dgn@chromium.org changed reviewers:
+ bauerb@chromium.org

[dgn, 2016-02-23 16:58:46]

PTAL:

TBD: tests, doc, update options in ScopedKeepAlive callers.

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
File chrome/browser/lifetime/keep_alive_options.h (right):

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_options.h:10: enum class
KeepAliveOptionRestart { DISABLED, ENABLED };
Should I have an extra member like UNKNOWN that would be invalid to make sure
callers don't use the default value?

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:32: void
KeepAliveRegistry::SetObserver(KeepAliveStateObserver* observer) {
having add/remove observer and an ObserverList would allow using a debug
observer for logging events.

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:52: KeepAliveOptions
KeepAliveRegistry::ComputeCurrentState() const {
Might be better to have a different struct instead of KeepAliveOptions. That
would allow skipping the label and having keeping_alive in the state.

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:80: KeepAliveOptions before_state
= ComputeCurrentState();
Should I store it in a member instead of recomputing? Wouldn't save much

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:122: msg +=
std::string(registered->label) + " ";
seems that we use std::string::append and base::StringPrintf for such operations

[dgn, 2016-02-24 13:03:56]

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:32: void
KeepAliveRegistry::SetObserver(KeepAliveStateObserver* observer) {
On 2016/02/23 16:58:46, dgn wrote:
> having add/remove observer and an ObserverList would allow using a debug
> observer for logging events.

Done.

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:52: KeepAliveOptions
KeepAliveRegistry::ComputeCurrentState() const {
On 2016/02/23 16:58:46, dgn wrote:
> Might be better to have a different struct instead of KeepAliveOptions. That
> would allow skipping the label and having keeping_alive in the state.

Done.

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:80: KeepAliveOptions before_state
= ComputeCurrentState();
On 2016/02/23 16:58:46, dgn wrote:
> Should I store it in a member instead of recomputing? Wouldn't save much

Done.

https://codereview.chromium.org/1725883002/diff/1/chrome/browser/lifetime/kee...
chrome/browser/lifetime/keep_alive_registry.cc:122: msg +=
std::string(registered->label) + " ";
On 2016/02/23 16:58:46, dgn wrote:
> seems that we use std::string::append and base::StringPrintf for such
operations

Just logged directly with streams

[Bernhard Bauer, 2016-02-24 16:54:34]

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/backgrou...
File chrome/browser/background/background_mode_manager.cc (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/backgrou...
chrome/browser/background/background_mode_manager.cc:314: new
ScopedKeepAlive(keep_alive::Origin::BACKGROUND_MODE_MANAGER_STARTUP,
Are we ever going to use the same origin with different restart options? If not,
it seems kind of strange to centralize origins themselves but keep the restart
options at the call sites. Maybe something to bring up  with Scott over on the
other CL?

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/backgrou...
chrome/browser/background/background_mode_manager.cc:769: if (!keep_alive_)
Keep the braces, as the body is more than one line.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:22: void HugBrowser() {
Awww... 😃

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:70: keeping_alive;
Hm... is restart really disallowed if nothing is keeping the browser alive? It
kind of feels like that isn't really well-defined in this case, so maybe would
it make more sense to make this a tri-state enum?

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:93: if (restart ==
keep_alive::RestartOption::ENABLED) {
Nit: No braces.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:109: if (restart ==
keep_alive::RestartOption::ENABLED) {
Nit: Braces are unnecessary here.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
File chrome/browser/lifetime/keep_alive_registry.h (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.h:29: struct KeepAliveState {
Do you need this as a struct? It seems a bit easier to just split this up into
two members / parameters.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.h:31: keep_alive::RestartOption
restart;
I'd be okay with keeping this as a boolean; I only want an enum for types that
are used as part of the public interface, but internally it's easier to keep
track of the interpretation of a boolean flag.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.h:53: KeepAliveState state_;
Hm... so, this is really defined by the two sets above. I think we could remove
this member in favor of always computing it from the canonical source. For state
transitions, you would then:

1) Calculate the old state, store locally
2) Modify state
3) Calculate the new state, store locally
4) Find out diff between old state and new state and notify observers about it.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
File chrome/browser/lifetime/keep_alive_types.cc (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_types.cc:39: default:
This isn't necessary if you enumerate all the values. You can move this out of
the switch statement and add a NOTREACHED().

[dgn, 2016-02-24 19:08:32]

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/backgrou...
File chrome/browser/background/background_mode_manager.cc (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/backgrou...
chrome/browser/background/background_mode_manager.cc:314: new
ScopedKeepAlive(keep_alive::Origin::BACKGROUND_MODE_MANAGER_STARTUP,
On 2016/02/24 16:54:33, Bernhard Bauer wrote:
> Are we ever going to use the same origin with different restart options? If
not,
> it seems kind of strange to centralize origins themselves but keep the restart
> options at the call sites. Maybe something to bring up  with Scott over on the
> other CL?

Up to callers to decide what to do with their origins I guess? My interpretation
is that since we don't have strings to declare and risk typos, spreading enums
is less of an issue. So in a sense it's more consistent with the changes Scott
had me make.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:70: keeping_alive;
On 2016/02/24 16:54:33, Bernhard Bauer wrote:
> Hm... is restart really disallowed if nothing is keeping the browser alive? It
> kind of feels like that isn't really well-defined in this case, so maybe would
> it make more sense to make this a tri-state enum?

Hum true... The observer can always check WillKeepAlive() if it cares. The
boolean still works then, and state change checking is just as easy.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
File chrome/browser/lifetime/keep_alive_registry.h (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.h:29: struct KeepAliveState {
On 2016/02/24 16:54:34, Bernhard Bauer wrote:
> Do you need this as a struct? It seems a bit easier to just split this up into
> two members / parameters.

Makes it easier to use compute/compare state

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.h:31: keep_alive::RestartOption
restart;
On 2016/02/24 16:54:34, Bernhard Bauer wrote:
> I'd be okay with keeping this as a boolean; I only want an enum for types that
> are used as part of the public interface, but internally it's easier to keep
> track of the interpretation of a boolean flag.

Thanks, I did it initially because I was comparing with KeepAliveOptions (RIP).
Not needed anymore.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.h:53: KeepAliveState state_;
On 2016/02/24 16:54:34, Bernhard Bauer wrote:
> Hm... so, this is really defined by the two sets above. I think we could
remove
> this member in favor of always computing it from the canonical source. For
state
> transitions, you would then:
> 
> 1) Calculate the old state, store locally
> 2) Modify state
> 3) Calculate the new state, store locally
> 4) Find out diff between old state and new state and notify observers about
it.

Ok, I did that way in PS1 and ended up storing it to save a calculation.
Changing..

[dgn, 2016-02-26 14:34:29]

PTAL

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:22: void HugBrowser() {
On 2016/02/24 16:54:34, Bernhard Bauer wrote:
> Awww... 😃

Aaaand.. It's gone.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:93: if (restart ==
keep_alive::RestartOption::ENABLED) {
On 2016/02/24 16:54:34, Bernhard Bauer wrote:
> Nit: No braces.

Removed.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_registry.cc:109: if (restart ==
keep_alive::RestartOption::ENABLED) {
On 2016/02/24 16:54:34, Bernhard Bauer wrote:
> Nit: Braces are unnecessary here.

Removed.

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
File chrome/browser/lifetime/keep_alive_types.cc (right):

https://codereview.chromium.org/1725883002/diff/20001/chrome/browser/lifetime...
chrome/browser/lifetime/keep_alive_types.cc:39: default:
On 2016/02/24 16:54:34, Bernhard Bauer wrote:
> This isn't necessary if you enumerate all the values. You can move this out of
> the switch statement and add a NOTREACHED().

Done.

[dgn, 2016-02-26 18:44:36]

Description was changed from

==========
[Draft] Add KeepAliveStateObserver, add the Restart option

BUG=587926,585080
==========

to

==========
Add KeepAliveStateObserver, add the Restart option

BUG=587926,585080
==========

[dgn, 2016-02-26 19:00:47]

Description was changed from

==========
Add KeepAliveStateObserver, add the Restart option

BUG=587926,585080
==========

to

==========
Add KeepAliveStateObserver, that allows clients to get
notified of state changes in the KeepAliveRegistry.
The first supported option is KeepAliveRestartOption.

This CL does not add any observer to act on that yet.

BUG=587926,585080
==========

[dgn, 2016-02-26 19:17:29]

dgn@chromium.org changed reviewers:
+ sky@chromium.org

[dgn, 2016-02-26 19:17:29]

PTAL

[sky, 2016-02-26 22:13:25]

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.cc:49: bool old_keeping_alive =
IsKeepingAlive();
49-50 and 58-65 are duplicated in UnRegister. How about a helper class whose
constructor has 49-50 and destructor 58-65?

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.h (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.h:56: int restart_allowed_count_;
Add description.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry_unittest.cc (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry_unittest.cc:19: registry_ =
KeepAliveRegistry::GetInstance();
nit: use member initialize.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry_unittest.cc:37: int
on_restart_allowed_call_count_;
nit: protected.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_state_observer.h (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_state_observer.h:14: virtual void
OnRestartAllowed() = 0;
How about an OnKeepAliveRestartStateChanged(bool can_restart);

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_state_observer.h:18: KeepAliveStateObserver()
{}
nit: I don't think you need this.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_state_observer.h:21:
DISALLOW_COPY_AND_ASSIGN(KeepAliveStateObserver);
you shouldn't need this as KeepAliveStateObserver has pure virtual functions.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/scoped_keep_alive.h (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/scoped_keep_alive.h:17: explicit
ScopedKeepAlive(KeepAliveOrigin origin,
nit: no explicit.

[dgn, 2016-03-07 17:17:14]

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.cc:49: bool old_keeping_alive =
IsKeepingAlive();
On 2016/02/26 22:13:25, sky wrote:
> 49-50 and 58-65 are duplicated in UnRegister. How about a helper class whose
> constructor has 49-50 and destructor 58-65?

I considered something similar before (see up to patchset 4 for example:
https://codereview.chromium.org/1725883002/diff2/60001:100001/chrome/browser/...)

An issue is that we might want to revert the notification order between adding
and removing, in case there are dependencies. Here the KeepAlive and
RestartAllowed notifications are swapped for example. I made quite granular
notifying and state reading methods to allow for that while keeping it simple
and concise in Register/Unregister. So you prefer the approach from patch 4?

[dgn, 2016-03-09 16:35:32]

PTAL

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.h (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.h:56: int restart_allowed_count_;
On 2016/02/26 22:13:25, sky wrote:
> Add description.

Done.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry_unittest.cc (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry_unittest.cc:19: registry_ =
KeepAliveRegistry::GetInstance();
On 2016/02/26 22:13:25, sky wrote:
> nit: use member initialize.

Done.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry_unittest.cc:37: int
on_restart_allowed_call_count_;
On 2016/02/26 22:13:25, sky wrote:
> nit: protected.

Done.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_state_observer.h (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_state_observer.h:14: virtual void
OnRestartAllowed() = 0;
On 2016/02/26 22:13:25, sky wrote:
> How about an OnKeepAliveRestartStateChanged(bool can_restart);

Yes done. I started with split methods as my current implementations of them
were very different, but that's a detail in the implementation, and should not
be relevant to the interface here.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_state_observer.h:18: KeepAliveStateObserver()
{}
On 2016/02/26 22:13:25, sky wrote:
> nit: I don't think you need this.

Removed.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_state_observer.h:21:
DISALLOW_COPY_AND_ASSIGN(KeepAliveStateObserver);
On 2016/02/26 22:13:25, sky wrote:
> you shouldn't need this as KeepAliveStateObserver has pure virtual functions.

Done.

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/scoped_keep_alive.h (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/scoped_keep_alive.h:17: explicit
ScopedKeepAlive(KeepAliveOrigin origin,
On 2016/02/26 22:13:25, sky wrote:
> nit: no explicit.

Done.

[sky, 2016-03-09 18:29:35]

LGTM

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.cc:49: bool old_keeping_alive =
IsKeepingAlive();
On 2016/03/07 17:17:14, dgn wrote:
> On 2016/02/26 22:13:25, sky wrote:
> > 49-50 and 58-65 are duplicated in UnRegister. How about a helper class whose
> > constructor has 49-50 and destructor 58-65?
> 
> I considered something similar before (see up to patchset 4 for example:
>
https://codereview.chromium.org/1725883002/diff2/60001:100001/chrome/browser/...)
> 
> An issue is that we might want to revert the notification order between adding
> and removing, in case there are dependencies. Here the KeepAlive and
> RestartAllowed notifications are swapped for example. I made quite granular
> notifying and state reading methods to allow for that while keeping it simple
> and concise in Register/Unregister. So you prefer the approach from patch 4?

Ok. Keep what you have here.

https://codereview.chromium.org/1725883002/diff/120001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.h (right):

https://codereview.chromium.org/1725883002/diff/120001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.h:57: // Number of registered keep
alives that have KeepAliveRestartOption::ENABLED
nit: end with '.'.

[dgn, 2016-03-09 18:37:04]

Thanks!

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.cc (right):

https://codereview.chromium.org/1725883002/diff/100001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.cc:49: bool old_keeping_alive =
IsKeepingAlive();
On 2016/03/09 18:29:35, sky wrote:
> On 2016/03/07 17:17:14, dgn wrote:
> > On 2016/02/26 22:13:25, sky wrote:
> > > 49-50 and 58-65 are duplicated in UnRegister. How about a helper class
whose
> > > constructor has 49-50 and destructor 58-65?
> > 
> > I considered something similar before (see up to patchset 4 for example:
> >
>
https://codereview.chromium.org/1725883002/diff2/60001:100001/chrome/browser/...)
> > 
> > An issue is that we might want to revert the notification order between
adding
> > and removing, in case there are dependencies. Here the KeepAlive and
> > RestartAllowed notifications are swapped for example. I made quite granular
> > notifying and state reading methods to allow for that while keeping it
simple
> > and concise in Register/Unregister. So you prefer the approach from patch 4?
> 
> Ok. Keep what you have here.

Acknowledged.

https://codereview.chromium.org/1725883002/diff/120001/chrome/browser/lifetim...
File chrome/browser/lifetime/keep_alive_registry.h (right):

https://codereview.chromium.org/1725883002/diff/120001/chrome/browser/lifetim...
chrome/browser/lifetime/keep_alive_registry.h:57: // Number of registered keep
alives that have KeepAliveRestartOption::ENABLED
On 2016/03/09 18:29:35, sky wrote:
> nit: end with '.'.

Done.

[dgn, 2016-03-09 18:37:10]

The CQ bit was checked by dgn@chromium.org

[dgn, 2016-03-09 18:37:10]

The patchset sent to the CQ was uploaded after l-g-t-m from sky@chromium.org
Link to the patchset: https://codereview.chromium.org/1725883002/#ps140001
(title: "Address comments")

[commit-bot: I haz the power, 2016-03-09 18:37:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1725883002/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1725883002/140001

[commit-bot: I haz the power, 2016-03-09 19:28:34]

Description was changed from

==========
Add KeepAliveStateObserver, that allows clients to get
notified of state changes in the KeepAliveRegistry.
The first supported option is KeepAliveRestartOption.

This CL does not add any observer to act on that yet.

BUG=587926,585080
==========

to

==========
Add KeepAliveStateObserver, that allows clients to get
notified of state changes in the KeepAliveRegistry.
The first supported option is KeepAliveRestartOption.

This CL does not add any observer to act on that yet.

BUG=587926,585080
==========

[commit-bot: I haz the power, 2016-03-09 19:28:35]

Committed patchset #8 (id:140001)

[commit-bot: I haz the power, 2016-03-09 19:30:44]

Description was changed from

==========
Add KeepAliveStateObserver, that allows clients to get
notified of state changes in the KeepAliveRegistry.
The first supported option is KeepAliveRestartOption.

This CL does not add any observer to act on that yet.

BUG=587926,585080
==========

to

==========
Add KeepAliveStateObserver, that allows clients to get
notified of state changes in the KeepAliveRegistry.
The first supported option is KeepAliveRestartOption.

This CL does not add any observer to act on that yet.

BUG=587926,585080

Committed: https://crrev.com/3563ecaf39e0c08836a17743d5233de1e0e7b6f0
Cr-Commit-Position: refs/heads/master@{#380174}
==========

[commit-bot: I haz the power, 2016-03-09 19:30:45]

Patchset 8 (id:??) landed as
https://crrev.com/3563ecaf39e0c08836a17743d5233de1e0e7b6f0
Cr-Commit-Position: refs/heads/master@{#380174}