Move ServiceWorkerCache backend creation to a lazy init function.

Move ServiceWorkerCache backend creation to a lazy init function.

Previously the ServiceWorkerCacheStorage called ServiceWorkerCache::CreateBackend on Get() and Create().  Now the ServiceWorkerCache does it itself lazily.  This is simpler for the ServiceWorkerCache's caller and should improve performance as backends are created only when needed.  This makes the downstream refcounting CL simpler.

Related CLs:
1. https://crrev.com/542703002: Change ownership of the parameters to ServiceWorkerCache:: Put and Match.
* 2. https://crrev.com/545533002: Move ServiceWorkerCache backend creation to a lazy init function.
3. https://crrev.com/548533002: Make ServiceWorkerCacheStorage::CacheLoader::LoadCache synchronous
4. https://crrev.com/549493002: Expose ServiceWorkerCache objects to ServiceWorkerCacheStorageManager clients.

BUG=392621
Committed: https://crrev.com/655e378c7d9d07b3937967c428fd51353e154d53
Cr-Commit-Position: refs/heads/master@{#294372}

[jkarlin, 2014-09-05 16:09:17]

jkarlin@chromium.org changed reviewers:
+ michaeln@chromium.org

[jkarlin, 2014-09-05 16:09:17]

Please take a look at everything, thanks!

[michaeln1, 2014-09-05 23:28:48]

michaeln@google.com changed reviewers:
+ michaeln@google.com

[michaeln1, 2014-09-05 23:28:49]

lgtm, nice cleanup!

https://codereview.chromium.org/545533002/diff/20001/content/browser/service_...
File content/browser/service_worker/service_worker_cache_unittest.cc (right):

https://codereview.chromium.org/545533002/diff/20001/content/browser/service_...
content/browser/service_worker/service_worker_cache_unittest.cc:282: // The
BlobHandle needs to be held onto by the caller until either the
That's too bad.

Not suggesting this for this cl, but maybe it would be nice to add a
ServiceWorkerFetchResponse derivative that's for use in the browser-process-only
to own the blobdatahandle for the duration. Having consumers of the repsonse obj
manage that out-of-band of the resp object itself is a pita.

[jkarlin, 2014-09-09 19:14:34]

https://codereview.chromium.org/545533002/diff/20001/content/browser/service_...
File content/browser/service_worker/service_worker_cache_unittest.cc (right):

https://codereview.chromium.org/545533002/diff/20001/content/browser/service_...
content/browser/service_worker/service_worker_cache_unittest.cc:282: // The
BlobHandle needs to be held onto by the caller until either the
On 2014/09/05 23:28:49, michaeln1 wrote:
> That's too bad.
> 
> Not suggesting this for this cl, but maybe it would be nice to add a
> ServiceWorkerFetchResponse derivative that's for use in the
browser-process-only
> to own the blobdatahandle for the duration. Having consumers of the repsonse
obj
> manage that out-of-band of the resp object itself is a pita.

Done.

[michaeln1, 2014-09-09 22:01:00]

still lgtm!

[michaeln, 2014-09-11 00:47:22]

and lgtm 2

[jkarlin, 2014-09-11 10:56:39]

The CQ bit was checked by jkarlin@chromium.org

[commit-bot: I haz the power, 2014-09-11 10:56:51]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patchset/545533002/80001

[commit-bot: I haz the power, 2014-09-11 11:55:12]

Committed patchset #5 (id:80001) as 80ca9bd1135de341767369934d59435a45086ff6

[commit-bot: I haz the power, 2014-09-11 11:57:47]

Patchset 5 (id:??) landed as
https://crrev.com/655e378c7d9d07b3937967c428fd51353e154d53
Cr-Commit-Position: refs/heads/master@{#294372}