[sync] Move exponential backoff tests / infrastructure into a new file

[sync] Move exponential backoff tests / infrastructure into a new file

This is another in a series of CLs to clean up ProfileSyncServiceHarness.
It does the following:

1) Moves the sync integration tests for exponential backoff into their
   own file.
2) Moves the test infrastructure used solely by these tests out of
   ProfileSyncServiceHarness and into the test file.
3) Reduces StatusChangeChecker to an interface, and moves its callback
   mechanism into ProfileSyncServiceHarness, where it can be used with
   any arbitrary method of PSSH. Other one-off users of SCC aren't forced
   to implement a callback, and can directly override IsExitConditionSatisfied.

BUG=323380
TEST=sync_integration_tests:SyncExponentialBackoffTest*
[email protected]

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=241350

[Raghu Simha, 2013-12-14 00:51:19]

Richard, please review. Thanks.

[rlarocque, 2013-12-14 01:09:17]

Woo-hoo!  This is exactly the sort of change I was hoping to see once we had
broken the state machine.

I've still got a few comments, though.

https://codereview.chromium.org/109853004/diff/20001/chrome/browser/sync/test...
File chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc
(right):

https://codereview.chromium.org/109853004/diff/20001/chrome/browser/sync/test...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:30: :
StatusChangeChecker(
I don't think it makes sense to reuse the full StatusChangeChecker here.

There should be a common class higher up in the class hierarchy that both
StatusChangeChecker and ExponentialBackoffChecker could inherit from.

This would be preferable to playing tricks with base::Bind at run time and
passing around 'this' in a base::Unretained() pointer.  This object should know
its own type when IsExitConditionSatisfied() is called.

[Raghu Simha, 2013-12-14 02:10:16]

PTAL.

https://codereview.chromium.org/109853004/diff/20001/chrome/browser/sync/test...
File chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc
(right):

https://codereview.chromium.org/109853004/diff/20001/chrome/browser/sync/test...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:30: :
StatusChangeChecker(
On 2013/12/14 01:09:17, rlarocque wrote:
> There should be a common class higher up in the class hierarchy that both
> StatusChangeChecker and ExponentialBackoffChecker could inherit from.

Right. Makes sense. Done.

[rlarocque, 2013-12-16 18:16:00]

Here's a few more comments.  Mostly nits.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
File chrome/browser/sync/test/integration/profile_sync_service_harness.cc
(right):

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/profile_sync_service_harness.cc:114: class
SyncStatusChangeChecker : public StatusChangeChecker {
SyncStatusChangeChecker isn't a very good name.  All of the classes that extend
StatusChangeChecker are likely to be related to sync in some way.

Here are some ideas for alternative names:
- SimpleStatusChangeChecker
- CallbackStatusChangeChecker
- GenericStatusChangeChecker

None of these are great, either.  Other suggestions welcome.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/profile_sync_service_harness.cc:132: //
Callback that binds a PSSH object to any of its methods that can evaluate
Technically, that's not true.  This class does nothing to enforce the use of a
PSSH.  In fact, I'm hoping some variants of it might reference the PSS rather
than the PSSH, which might allow us to shrink PSSH even more than you've already
done.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
File chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc
(right):

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:27: class
ExponentialBackoffChecker : public StatusChangeChecker {
nit: This class could be placed in an anonymous namespace.

If these tests aren't relying on any FRIEND_TEST or other friend declarations,
you could move most of the contents of this file into an anonymous namespace.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:54:
scoped_ptr<RetryVerifier> retry_verifier_;
Does this need to be a scoped_ptr?  Why not make it a member so you don't need
to worry about initializing it?

[Raghu Simha, 2013-12-16 19:02:31]

Richard, PTAL.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
File chrome/browser/sync/test/integration/profile_sync_service_harness.cc
(right):

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/profile_sync_service_harness.cc:114: class
SyncStatusChangeChecker : public StatusChangeChecker {
On 2013/12/16 18:16:01, rlarocque wrote:
> SyncStatusChangeChecker isn't a very good name.  All of the classes that
extend
> StatusChangeChecker are likely to be related to sync in some way.
> 
> Here are some ideas for alternative names:
> - SimpleStatusChangeChecker
> - CallbackStatusChangeChecker
> - GenericStatusChangeChecker
> 
> None of these are great, either.  Other suggestions welcome.

I went with CallbackStatusChecker.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/profile_sync_service_harness.cc:132: //
Callback that binds a PSSH object to any of its methods that can evaluate
On 2013/12/16 18:16:01, rlarocque wrote:
> Technically, that's not true.  This class does nothing to enforce the use of a
> PSSH.  In fact, I'm hoping some variants of it might reference the PSS rather
> than the PSSH, which might allow us to shrink PSSH even more than you've
already
> done.

Agreed.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
File chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc
(right):

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:27: class
ExponentialBackoffChecker : public StatusChangeChecker {
On 2013/12/16 18:16:01, rlarocque wrote:
> nit: This class could be placed in an anonymous namespace.
> 
> If these tests aren't relying on any FRIEND_TEST or other friend declarations,
> you could move most of the contents of this file into an anonymous namespace.

Done.

https://codereview.chromium.org/109853004/diff/120001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:54:
scoped_ptr<RetryVerifier> retry_verifier_;
On 2013/12/16 18:16:01, rlarocque wrote:
> Does this need to be a scoped_ptr?  Why not make it a member so you don't need
> to worry about initializing it?

This was originally a scoped_ptr member of PSSH so that it couldn't be
inadvertently used more than once without being reinitialized. In this case, It
can be a member.

[rlarocque, 2013-12-16 20:01:03]

LGTM.

https://codereview.chromium.org/109853004/diff/140001/chrome/browser/sync/tes...
File chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc
(right):

https://codereview.chromium.org/109853004/diff/140001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:60: }  //
namespace
nit: It should be possible to include the tests themselves in this scope, right?

[Raghu Simha, 2013-12-16 20:43:21]

https://codereview.chromium.org/109853004/diff/140001/chrome/browser/sync/tes...
File chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc
(right):

https://codereview.chromium.org/109853004/diff/140001/chrome/browser/sync/tes...
chrome/browser/sync/test/integration/sync_exponential_backoff_test.cc:60: }  //
namespace
On 2013/12/16 20:01:03, rlarocque wrote:
> nit: It should be possible to include the tests themselves in this scope,
right?

Done.

[commit-bot: I haz the power, 2013-12-16 20:45:29]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/109853004/220001

[commit-bot: I haz the power, 2013-12-16 21:25:38]

Retried try job too often on mac_rel for step(s) base_unittests, browser_tests,
interactive_ui_tests, net_unittests, unit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...

[commit-bot: I haz the power, 2013-12-16 21:50:32]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/109853004/240001

[Raghu Simha, 2013-12-17 20:41:27]

Committed patchset #6 manually as r241350 (presubmit successful).