Allow showing pending URL for new tab navigations, but only if safe.

Allow showing pending URL for new tab navigations, but only if safe.

We revert to showing the opener's URL if it modifies the content of the
initial blank page before the pending URL commits, to prevent URL spoofs.

Implications:
 - All renderer-initiated navigations now have pending NavigationEntries.
 - GetURL and GetTitle use the visible entry, not active entry.
 - The renderer notifies the browser if DOM mutations occur before first load.

BUG=9682
TEST=Open a slow link in a new tab. Destination URL should be visible and reloadable.


Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=191277

[Charlie Reis, 2013-03-07 17:28:20]

Hey Darin--
  This CL to fix 9682 should be passing all tests except for
FormPostBackForward_0 on ChromeFrame at this point.  I'll try to talk more with
the ChromeFrame folks about that one.

  Let me know if you'd like me to break this into multiple CLs, since I may be
able to separate out the pending_entry stuff from the DidAccessInitialDocument
stuff.  I'm somewhat inclined to leave it as one CL to make the motivation
clearer, but I can split it up if that would help.

Thanks!
Charlie

[Charlie Reis, 2013-03-07 20:36:42]

FYI: I've split out the IsInitialNavigation change into
https://codereview.chromium.org/12637002/.

[darin (slow to review), 2013-03-10 05:11:06]

LGTM

[grt (UTC plus 2), 2013-03-11 18:31:14]

The change appears to do what Robert explained to me, but I don't have the depth
in network-land to know if there are cases where this change will do the wrong
thing.

[ananta, 2013-03-13 18:01:11]

lgtm

[Charlie Reis, 2013-03-21 01:05:57]

Tom, can you review view_messages.h?  Just one new zero-arg IPC.

[Tom Sepez, 2013-03-21 17:12:41]

On 2013/03/21 01:05:57, creis wrote:
> Tom, can you review view_messages.h?  Just one new zero-arg IPC.

LGTM

[joth, 2013-03-22 17:12:52]

lgtm

https://codereview.chromium.org/12541018/diff/50001/content/browser/android/c...
File content/browser/android/content_view_core_impl.cc (right):

https://codereview.chromium.org/12541018/diff/50001/content/browser/android/c...
content/browser/android/content_view_core_impl.cc:751: content::NavigationEntry*
entry =
I think this should comment why it's not using GetURL() (even if just saying the
change from crrev.com/12541018 was incompatibile with the existing users of the
Java API)

[Charlie Reis, 2013-03-28 19:23:19]

Darin, can you take a look at the delta between patch sets 8 and 9?  It's a
small change WebContentsImpl::GetTitle and navigation_controller_unittest.cc.

The CFInspector test turned out to be a real issue, where we were showing the
pending entry's URL in the title when navigating a new tab.  It's supposed to be
the empty string, which gets converted to "Loading...".  Fixed by only doing the
Ctrl+Back check if GetLastCommittedEntry exists.

With that, I think it passes all the tests that I know of.

https://codereview.chromium.org/12541018/diff/50001/content/browser/android/c...
File content/browser/android/content_view_core_impl.cc (right):

https://codereview.chromium.org/12541018/diff/50001/content/browser/android/c...
content/browser/android/content_view_core_impl.cc:751: content::NavigationEntry*
entry =
On 2013/03/22 17:12:52, joth wrote:
> I think this should comment why it's not using GetURL() (even if just saying
the
> change from crrev.com/12541018 was incompatibile with the existing users of
the
> Java API)

Done.

[darin (slow to review), 2013-03-28 21:45:33]

OK, LGTM

[darin (slow to review), 2013-03-28 21:45:34]

OK, LGTM

[commit-bot: I haz the power, 2013-03-29 00:28:13]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/creis@chromium.org/12541018/70001

[commit-bot: I haz the power, 2013-03-29 03:29:05]

Change committed as 191277