Add extra data to BrowserContext so that content layer and other embedders can stash data with it t…

Add extra data to BrowserContext so that content layer and other embedders can stash data with it that has the same lifetime. Converted SSLHostState to use it for now. I'll do the rest in a followup.

BUG=98716
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=122164

[jam, 2012-02-14 22:08:59]

here is change we talked about

[darin (slow to review), 2012-02-15 00:54:42]

http://codereview.chromium.org/9348109/diff/4014/content/public/browser/brows...
File content/public/browser/browser_context.h (right):

http://codereview.chromium.org/9348109/diff/4014/content/public/browser/brows...
content/public/browser/browser_context.h:50: class UserData {
Are you sure you want a map?  It seems like you could get by permitting only one
UserData per BrowserContext.

Also, I wonder if it would be useful to define a class named
base::SupportsUserData that URLRequest and BrowserContext could each derive from
(assuming we wanted to keep the map).  Interesting idea?

[jam, 2012-02-15 01:01:17]

http://codereview.chromium.org/9348109/diff/4014/content/public/browser/brows...
File content/public/browser/browser_context.h (right):

http://codereview.chromium.org/9348109/diff/4014/content/public/browser/brows...
content/public/browser/browser_context.h:50: class UserData {
On 2012/02/15 00:54:42, darin wrote:
> Are you sure you want a map?  It seems like you could get by permitting only
one
> UserData per BrowserContext.

I was thinking that each object that hangs off BrowserContext could have its own
object.
if we only have one, we probably don't want to expose this to embedders then?

> 
> Also, I wonder if it would be useful to define a class named
> base::SupportsUserData that URLRequest and BrowserContext could each derive
from
> (assuming we wanted to keep the map).  Interesting idea?

i like this, it'll probably make other code cleaner too.

[darin (slow to review), 2012-02-15 06:19:05]

http://codereview.chromium.org/9348109/diff/4042/base/supports_user_data.h
File base/supports_user_data.h (right):

http://codereview.chromium.org/9348109/diff/4042/base/supports_user_data.h#ne...
base/supports_user_data.h:24: class BASE_EXPORT Data {
curious: why drop the "User" prefix?

http://codereview.chromium.org/9348109/diff/4042/content/browser/renderer_hos...
File content/browser/renderer_host/resource_dispatcher_host_unittest.cc (left):

http://codereview.chromium.org/9348109/diff/4042/content/browser/renderer_hos...
content/browser/renderer_host/resource_dispatcher_host_unittest.cc:294: void
set_url_request_user_data(net::URLRequest::UserData* user_data) {
shouldn't net::URLRequest::UserData still work even if UserData is defined on
the base class of URLRequest?

[jam, 2012-02-15 15:44:42]

http://codereview.chromium.org/9348109/diff/4042/base/supports_user_data.h
File base/supports_user_data.h (right):

http://codereview.chromium.org/9348109/diff/4042/base/supports_user_data.h#ne...
base/supports_user_data.h:24: class BASE_EXPORT Data {
On 2012/02/15 06:19:05, darin wrote:
> curious: why drop the "User" prefix?

before, it used to be URLRequest::UserData which was descriptive. now, it's
SupportsUserData::Data which I think is descriptive as well.
SupportsUserData::UserData seemed redundant

http://codereview.chromium.org/9348109/diff/4042/content/browser/renderer_hos...
File content/browser/renderer_host/resource_dispatcher_host_unittest.cc (left):

http://codereview.chromium.org/9348109/diff/4042/content/browser/renderer_hos...
content/browser/renderer_host/resource_dispatcher_host_unittest.cc:294: void
set_url_request_user_data(net::URLRequest::UserData* user_data) {
On 2012/02/15 06:19:05, darin wrote:
> shouldn't net::URLRequest::UserData still work even if UserData is defined on
> the base class of URLRequest?

I removed the net::URLRequest::UserData class, since there's no need for it.

[joth, 2012-08-23 18:11:52]

https://chromiumcodereview.appspot.com/9348109/diff/9002/base/supports_user_d...
File base/supports_user_data.h (right):

https://chromiumcodereview.appspot.com/9348109/diff/9002/base/supports_user_d...
base/supports_user_data.h:16: // key. At destruction all the objects will be
destructed.
Is there some guideline when to use this vs. when to use base::PropertyBag ?
they both seem to be targeted at exactly the same use-case.