Using WebFrame::createAssociatedURLLoader() instead of WebFrame::dispatchWillSendRequest(...)

Using WebFrame::createAssociatedURLLoader() instead of WebFrame::dispatchWillSendRequest(...)

According to,
http://codesearch.google.com/codesearch/p?hl=en#OAMlx_jo-ck/src/third_party/WebKit/WebKit/chromium/public/WebFrame.h
WebFrame::dispatchWillSendRequest is now deprecated.  Updating to use createAssociatedURLLoader() instead.

BUG=65806
TEST=media

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=69083

[scherkus (not reviewing), 2010-12-07 19:56:44]

do we have a corresponding bug and test?

other than that I think this is LGTM (although I'm not totally familiar with
createAssociatedURLLoader() -- i'll do some reading)

[annacc, 2010-12-08 00:03:52]

Thanks Andrew.  I've added a corresponding bug and I believe general media tests
should be sufficient 
(webkit/tools/layout_tests/run_webkit_tests.sh --debug http/tests/media). 

On 2010/12/07 19:56:44, scherkus wrote:
> do we have a corresponding bug and test?
> 
> other than that I think this is LGTM (although I'm not totally familiar with
> createAssociatedURLLoader() -- i'll do some reading)

[scherkus (not reviewing), 2010-12-08 01:20:49]

LGTM!

[annacc, 2010-12-09 20:25:56]

Thanks!  I'll go ahead and commit if it looks ok to hclam and darin.

On 2010/12/08 01:20:49, scherkus wrote:
> LGTM!

[michaeln, 2010-12-11 01:16:02]

Ah... i see... i'll drop my CL then :)

Could you also be sure to set the 'targetType' so its clear in the backend that
this is a media request, request.setTargetType(WebURLRequest::TargetIsMedia)...
look here for an example...
http://codereview.chromium.org/5765002/diff/1/webkit/glue/media/buffered_data...

That's needed so the browser knows to utilize ricardo's special sparse 'media
http cache' when servicing these requests. That should definitely be set for the
'buffered_data_source' since it's doing range requests, i'm not clear if that
should also be applied to the 'simple_data_source'? If 'simple_data_source' used
to use MediaResourceLoaderBridgeFactory, then to retain the old behavior, it too
should setTargetType() to MEDIA. The logic in the browser that looks at the
target type is...

URLRequestContext* ResourceMessageFilter::GetRequestContext(
    uint32 request_id,
    const ViewHostMsg_Resource_Request& request_data) {
  URLRequestContextGetter* request_context = request_context_;
  // If the request has resource type of ResourceType::MEDIA, we use a request
  // context specific to media for handling it because these resources have
  // specific needs for caching.
  if (request_data.resource_type == ResourceType::MEDIA)
    request_context = media_request_context_;
  return request_context->GetURLRequestContext();
}

[annacc, 2010-12-11 21:45:20]

Thanks michaeln!  I went ahead and added the setTargetType to requests in both
buffered_ and simple_ data_source.

[annacc, 2010-12-13 20:22:02]

Shall I go ahead and commit?

On 2010/12/11 21:45:20, annacc wrote:
> Thanks michaeln!  I went ahead and added the setTargetType to requests in both
> buffered_ and simple_ data_source.

[michaeln, 2010-12-13 21:35:32]

This LGTM!

[darin (slow to review), 2010-12-14 00:10:48]

LGTM2!

[annacc, 2010-12-14 00:57:57]

committed, revision 69083.