Switch ContentSettingsObserver to be a RenderFrameObserver instead of a RenderViewObserver

Switch ContentSettingsObserver to be a RenderFrameObserver instead of a RenderViewObserver

BUG=304341
[email protected], [email protected]

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=241306

[mkosiba (inactive), 2013-12-16 10:18:00]

https://codereview.chromium.org/104833006/diff/20001/android_webview/android_...
File android_webview/android_webview.gyp (right):

https://codereview.chromium.org/104833006/diff/20001/android_webview/android_...
android_webview/android_webview.gyp:214: 'renderer/aw_permission_client.h',
I think you forgot to add these to the CL

[jam, 2013-12-16 18:52:30]

https://codereview.chromium.org/104833006/diff/20001/android_webview/android_...
File android_webview/android_webview.gyp (right):

https://codereview.chromium.org/104833006/diff/20001/android_webview/android_...
android_webview/android_webview.gyp:214: 'renderer/aw_permission_client.h',
On 2013/12/16 10:18:01, mkosiba wrote:
> I think you forgot to add these to the CL

yep, thanks I've fixed that now. I wasn't able to do try runs until blink
rolled, but then I gave up and added a temporary roll in my change to be able to
test other platforms.

[nasko, 2013-12-16 19:47:36]

Just a few nits and some comments.

https://codereview.chromium.org/104833006/diff/130001/chrome/renderer/chrome_...
File chrome/renderer/chrome_content_renderer_client.cc (right):

https://codereview.chromium.org/104833006/diff/130001/chrome/renderer/chrome_...
chrome/renderer/chrome_content_renderer_client.cc:543: // TODO(jam): switch
ContentSettingsObserver to RenderFrameObserver.
nit: Looks like this is a stale comment at this point in time.

https://codereview.chromium.org/104833006/diff/130001/content/browser/frame_h...
File content/browser/frame_host/render_frame_host_delegate.h (right):

https://codereview.chromium.org/104833006/diff/130001/content/browser/frame_h...
content/browser/frame_host/render_frame_host_delegate.h:33: // not a
WebContents, returns NULL.
RenderViewHostDelegate has(had?) identical method that was marked deprecated and
not to be used. Should we be adding this if we wanted to remove the other such
method?

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:291: rfh->Send(message_copy);
Wouldn't this leak the message_copy? We allocate copy here, but don't free it.

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:336: WebContents*
WebContents::FromRenderFrameHost(RenderFrameHost* rfh) {
nit: Add // static prior to the method.

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:663: delete message;
nit: It will be useful to document in the header that the callee cannot use the
message after the call, since it is deleted.

https://codereview.chromium.org/104833006/diff/130001/content/public/browser/...
File content/public/browser/render_frame_host.h (right):

https://codereview.chromium.org/104833006/diff/130001/content/public/browser/...
content/public/browser/render_frame_host.h:26: virtual RenderProcessHost*
GetProcess() = 0;
Should be a const method.

https://codereview.chromium.org/104833006/diff/130001/content/renderer/render...
File content/renderer/render_frame_impl.h (right):

https://codereview.chromium.org/104833006/diff/130001/content/renderer/render...
content/renderer/render_frame_impl.h:67: void
MainWebFrameCreated(blink::WebFrame* frame);
Will we need this longer term? If not, might be useful to put a TODO to ensure
we don't miss it later on.

[jam, 2013-12-16 20:26:47]

https://codereview.chromium.org/104833006/diff/130001/chrome/renderer/chrome_...
File chrome/renderer/chrome_content_renderer_client.cc (right):

https://codereview.chromium.org/104833006/diff/130001/chrome/renderer/chrome_...
chrome/renderer/chrome_content_renderer_client.cc:543: // TODO(jam): switch
ContentSettingsObserver to RenderFrameObserver.
On 2013/12/16 19:47:37, nasko wrote:
> nit: Looks like this is a stale comment at this point in time.

Done.

https://codereview.chromium.org/104833006/diff/130001/content/browser/frame_h...
File content/browser/frame_host/render_frame_host_delegate.h (right):

https://codereview.chromium.org/104833006/diff/130001/content/browser/frame_h...
content/browser/frame_host/render_frame_host_delegate.h:33: // not a
WebContents, returns NULL.
On 2013/12/16 19:47:37, nasko wrote:
> RenderViewHostDelegate has(had?) identical method that was marked deprecated
and
> not to be used. Should we be adding this if we wanted to remove the other such
> method?

I had added that comment, but afterwards realized that it's not possible to
remove.

As for this method, I had started a discussion about this on site-isolation's
mailing list but it turned private between ajwong and me at the end; where he
saw that it was necessary to have this method. The reason is that we can
sometimes be on a non-UI thread with only a render_frame and render_process IDs,
and we need to get to a WebContents. as in this change. In that case, there's no
other way to get to WC except through the RF's delegate

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
File content/browser/web_contents/web_contents_impl.cc (right):

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:291: rfh->Send(message_copy);
On 2013/12/16 19:47:37, nasko wrote:
> Wouldn't this leak the message_copy? We allocate copy here, but don't free it.

An IPC is owned by the Send call. the original IPC is deleted in
WebContentsImpl::SendToAllFrames

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:336: WebContents*
WebContents::FromRenderFrameHost(RenderFrameHost* rfh) {
On 2013/12/16 19:47:37, nasko wrote:
> nit: Add // static prior to the method.

That's a personal preference that some people like. Myself, I don't like it
because it's documenting something that is trivially seen in the header. note
also the static methods above don' have this.

https://codereview.chromium.org/104833006/diff/130001/content/browser/web_con...
content/browser/web_contents/web_contents_impl.cc:663: delete message;
On 2013/12/16 19:47:37, nasko wrote:
> nit: It will be useful to document in the header that the callee cannot use
the
> message after the call, since it is deleted.

that is the convention for all methods that send an IPC...

https://codereview.chromium.org/104833006/diff/130001/content/public/browser/...
File content/public/browser/render_frame_host.h (right):

https://codereview.chromium.org/104833006/diff/130001/content/public/browser/...
content/public/browser/render_frame_host.h:26: virtual RenderProcessHost*
GetProcess() = 0;
On 2013/12/16 19:47:37, nasko wrote:
> Should be a const method.

I have avoided putting const on public interfaces in the content api because it
leaks implementation details into the public api.

https://codereview.chromium.org/104833006/diff/130001/content/renderer/render...
File content/renderer/render_frame_impl.h (right):

https://codereview.chromium.org/104833006/diff/130001/content/renderer/render...
content/renderer/render_frame_impl.h:67: void
MainWebFrameCreated(blink::WebFrame* frame);
On 2013/12/16 19:47:37, nasko wrote:
> Will we need this longer term? If not, might be useful to put a TODO to ensure
> we don't miss it later on.

this will be needed because when the RenderFrame is created, we don't have a
WebFrame yet. So it needs to be told so that it can tell its observers. this is
independent of whether a RF is per WF or per many.

[nasko, 2013-12-16 21:08:56]

LGTM

[jam, 2013-12-16 23:52:35]

+boliu for android_webview

[boliu, 2013-12-17 00:18:16]

aw lgtm