Tell the AppCacheService which request context to use when fetching resources...

1) Tell the AppCacheService which request context to use when fetching resources for updates. Done for both chrome and test_shell. The service does not yet take a reference to that context, because the extra reference apparently gives some tests grief.

2) Added methods to generate new storage ids for different object types on the IO thread.

BUG=none
TEST=none


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=26253

[michaeln, 2009-09-14 01:53:15]

A couple of small changes.

[jennb, 2009-09-14 20:43:25]

Thanks for adding this change.

http://codereview.chromium.org/195077/diff/6001/7002
File webkit/appcache/appcache_service.h (right):

http://codereview.chromium.org/195077/diff/6001/7002#newcode40
Line 40: // TODO(jennb): API to set service settings, like file paths for
storage
Can delete this TODO as you've already added filepath in Initialize().

http://codereview.chromium.org/195077/diff/6001/7002#newcode66
Line 66: // The service generates unique storage ids for differnt object types.
Are these impls only temporary Id generators with the real ids being generated
once we have a storage layer?  Do the ids need to be globally unique across all
services?

[michaeln, 2009-09-14 20:54:38]

I'm mystified by the try failures on windows still?

http://codereview.chromium.org/195077/diff/6001/7002
File webkit/appcache/appcache_service.h (right):

http://codereview.chromium.org/195077/diff/6001/7002#newcode40
Line 40: // TODO(jennb): API to set service settings, like file paths for
storage
On 2009/09/14 20:43:26, jennb wrote:
> Can delete this TODO as you've already added filepath in Initialize().

Done.

http://codereview.chromium.org/195077/diff/6001/7002#newcode66
Line 66: // The service generates unique storage ids for differnt object types.
On 2009/09/14 20:43:26, jennb wrote:
> Are these impls only temporary Id generators with the real ids being generated
> once we have a storage layer?  Do the ids need to be globally unique across
all
> services?

This is a real impl. An instance of a service has exclusive access to it's
database file. Other instance may exist, but they will be working with different
database files (for different profiles).

At init time, we'll determine what the last used ids are, and from that point
forward we can simply generate new ids at will.

[jennb, 2009-09-14 21:12:30]

http://codereview.chromium.org/195077/diff/6001/7002
File webkit/appcache/appcache_service.h (right):

http://codereview.chromium.org/195077/diff/6001/7002#newcode40
Line 40: // TODO(jennb): API to set service settings, like file paths for
storage
On 2009/09/14 20:54:38, michaeln wrote:
> On 2009/09/14 20:43:26, jennb wrote:
> > Can delete this TODO as you've already added filepath in Initialize().
> 
> Done.

The line is still here. :-)

http://codereview.chromium.org/195077/diff/6001/7002#newcode66
Line 66: // The service generates unique storage ids for differnt object types.
On 2009/09/14 20:54:38, michaeln wrote:
> On 2009/09/14 20:43:26, jennb wrote:
> > Are these impls only temporary Id generators with the real ids being
generated
> > once we have a storage layer?  Do the ids need to be globally unique across
> all
> > services?
> 
> This is a real impl. An instance of a service has exclusive access to it's
> database file. Other instance may exist, but they will be working with
different
> database files (for different profiles).
> 
> At init time, we'll determine what the last used ids are, and from that point
> forward we can simply generate new ids at will.
> 
> 

Can you add to the class comment about each service instance having its own db
file?

[michaeln, 2009-09-14 21:49:52]

> The line is still here. :-)

You looked too fast :) Its gone.

> Can you add to the class comment about each service instance having its own db
> file?

Done locally, i'm letting my tries run to completion before uploading a new
snapshot.

[michaeln, 2009-09-14 22:05:17]

There is try failures with this simple change that i have to pry into :(

[jennb, 2009-09-15 04:34:31]

LGTM

Looks like you figured out the trybot failure.  I'll sync this  CL before my
next snapshot, too, if you can commit early tomorrow.

Thanks!

[michaeln, 2009-09-15 17:03:18]

There are try failures in there still... the top level results are green... but
look inside.

[michaeln, 2009-09-15 18:23:09]

I'm working on committing this so the new AppCacheService methods are available
to call (request_context(), NextFooId()), but for now request_context() is
stubbed out to return NULL.