Schedule the stop-worker-timer even if the ServiceWorker is controlling pages

Schedule the stop-worker-timer even if the ServiceWorker is controlling pages

SSIA. I want to land this asap and see if it doesn't break anything...

BUG=440259
Committed: https://crrev.com/39937aec0cf59161ae5b2b4d57c1345a08120166
Cr-Commit-Position: refs/heads/master@{#308975}

[kinuko, 2014-12-18 07:18:30]

kinuko@chromium.org changed reviewers:
+ falken@chromium.org

[kinuko, 2014-12-18 07:20:04]

PTL.

[falken, 2014-12-18 07:28:54]

What happens if the SW is in the middle of handling an event when we try to kill
it?

I think ideally we'd let SW's handling an event or with ongoing work (unresolved
respondWith/waitUntil/cache.add promise etc) stay alive, but kill it if it is
running for too long. Some (maybe outdated) notes are on
https://github.com/slightlyoff/ServiceWorker/issues/107#issuecomment-32980998

[kinuko, 2014-12-18 07:59:46]

On 2014/12/18 07:28:54, falken wrote:
> What happens if the SW is in the middle of handling an event when we try to
kill
> it?
> 
> I think ideally we'd let SW's handling an event or with ongoing work
(unresolved
> respondWith/waitUntil/cache.add promise etc) stay alive, but kill it if it is
> running for too long. Some (maybe outdated) notes are on
> https://github.com/slightlyoff/ServiceWorker/issues/107#issuecomment-32980998

Currently we simply delay killing the ServiceWorker by rescheduling the timer
repeatedly.

For killing a ServiceWorker if an event handler is running too long, there's
another TODO noted here:
https://code.google.com/p/chromium/codesearch#chromium/src/content/browser/se...

[falken, 2014-12-18 08:26:40]

I see now, lgtm.

https://codereview.chromium.org/805083006/diff/1/content/browser/service_work...
File content/browser/service_worker/service_worker_version_unittest.cc (right):

https://codereview.chromium.org/805083006/diff/1/content/browser/service_work...
content/browser/service_worker/service_worker_version_unittest.cc:398:
EXPECT_GE(delay, version_->stop_worker_timer_.GetCurrentDelay());
I don't quite get this, what does GetCurrentDelay() return? Why does it get
smaller after the timer is reset?

[kinuko, 2014-12-18 08:31:41]

https://codereview.chromium.org/805083006/diff/1/content/browser/service_work...
File content/browser/service_worker/service_worker_version_unittest.cc (right):

https://codereview.chromium.org/805083006/diff/1/content/browser/service_work...
content/browser/service_worker/service_worker_version_unittest.cc:398:
EXPECT_GE(delay, version_->stop_worker_timer_.GetCurrentDelay());
On 2014/12/18 08:26:40, falken wrote:
> I don't quite get this, what does GetCurrentDelay() return? Why does it get
> smaller after the timer is reset?

Um, nvm, this line is not right (I wanted to compare scheduled_run_time_). I'm
removing this EXPECT line.

[kinuko, 2014-12-18 08:33:35]

The CQ bit was checked by kinuko@chromium.org

[commit-bot: I haz the power, 2014-12-18 08:34:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/805083006/20001

[commit-bot: I haz the power, 2014-12-18 09:18:07]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2014-12-18 09:18:46]

Patchset 2 (id:??) landed as
https://crrev.com/39937aec0cf59161ae5b2b4d57c1345a08120166
Cr-Commit-Position: refs/heads/master@{#308975}