Make content text retrieval in ChromeRenderViewObserver never force layout.

Make content text retrieval in ChromeRenderViewObserver never force layout.

Use the hooks, introduced in https://codereview.chromium.org/1309883003/
to remove the need for both searching for meta tags and an 8-second
speculative text retrieval on a timer.

This includes plumbing RenderViewObserver::DidFirstLayoutAfterFinishedParsing
so that the RenderView observer could us this call.

BUG=521149
R=jochen@chromium.org

[dglazkov, 2015-08-24 16:20:20]

dglazkov@chromium.org changed reviewers:
+ toyoshim@chromium.org

[dglazkov, 2015-08-24 16:20:42]

The CQ bit was checked by dglazkov@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-08-24 16:22:20]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1308113002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1308113002/20001

[commit-bot: I haz the power, 2015-08-24 16:58:40]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-08-24 16:58:41]

Dry run: Try jobs failed on following builders:
  mac_chromium_rel_ng on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)

[dglazkov, 2015-08-24 18:02:49]

The CQ bit was checked by dglazkov@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-08-24 18:04:14]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1308113002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1308113002/20001

[commit-bot: I haz the power, 2015-08-24 18:39:16]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-08-24 18:39:17]

Dry run: Try jobs failed on following builders:
  mac_chromium_rel_ng on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)

[nasko, 2015-08-24 22:59:05]

nasko@chromium.org changed reviewers:
+ nasko@chromium.org

[nasko, 2015-08-24 22:59:05]

https://codereview.chromium.org/1308113002/diff/20001/content/public/renderer...
File content/public/renderer/render_view_observer.h (right):

https://codereview.chromium.org/1308113002/diff/20001/content/public/renderer...
content/public/renderer/render_view_observer.h:62: virtual void
DidFirstLayoutAfterFinishedParsing() {}
We shouldn't be putting any new members on RenderViewObserver, as we are trying
to kill it. RenderFrameObserver should be used instead.

[dglazkov, 2015-08-25 02:56:46]

On 2015/08/24 at 22:59:05, nasko wrote:
>
https://codereview.chromium.org/1308113002/diff/20001/content/public/renderer...
> File content/public/renderer/render_view_observer.h (right):
> 
>
https://codereview.chromium.org/1308113002/diff/20001/content/public/renderer...
> content/public/renderer/render_view_observer.h:62: virtual void
DidFirstLayoutAfterFinishedParsing() {}
> We shouldn't be putting any new members on RenderViewObserver, as we are
trying to kill it. RenderFrameObserver should be used instead.

Interesting. RenderFrameObserver looks like a slightly odd abstraction for this.
All the hooks are all frame lifecycle-related. On the other hand, the
RenderViewObserver looks more like a view (compositor surface)-observing
abstraction. Why are you trying to kill it? It seems that we shouldn't be
conflating the two.

[dcheng, 2015-08-25 06:59:09]

On 2015/08/25 at 02:56:46, dglazkov wrote:
> On 2015/08/24 at 22:59:05, nasko wrote:
> >
https://codereview.chromium.org/1308113002/diff/20001/content/public/renderer...
> > File content/public/renderer/render_view_observer.h (right):
> > 
> >
https://codereview.chromium.org/1308113002/diff/20001/content/public/renderer...
> > content/public/renderer/render_view_observer.h:62: virtual void
DidFirstLayoutAfterFinishedParsing() {}
> > We shouldn't be putting any new members on RenderViewObserver, as we are
trying to kill it. RenderFrameObserver should be used instead.
> 
> Interesting. RenderFrameObserver looks like a slightly odd abstraction for
this. All the hooks are all frame lifecycle-related. On the other hand, the
RenderViewObserver looks more like a view (compositor surface)-observing
abstraction. Why are you trying to kill it? It seems that we shouldn't be
conflating the two.

A RenderView/WebView can have a remote main frame. In that case, there's no
compositor surface associated with the RenderView/WebView.

Quoting from the Blink review (since commenting on closed reviews is annoying):
> It's currently only called on main frame, because it's actually tied to
producing a frame (in the rendering sense -- the WebViewImpl::layout), rather
than just layout (in the document layout sense -- the FrameView::layout). I
don't have a better name for that, but maybe we should invent something.

Methods on WebViewClient should still make sense to call even if the WebView has
a remote main frame. It sounds like this method wouldn't make sense to call on a
WebView with a remote main frame (as a remote main frame doesn't perform
layout).