Clean up the way top level navigations are handled in ChromeFrame and reporte...

Clean up the way top level navigations are handled in ChromeFrame and reported back
to the host browser. We no longer have a counter which tracks the last page id.

Top level navigations are reported back if the current frame does not have an opener
or if the origins don't match. The other conditions remain the same i.e reloads/back forwards
and posts don't make it to the host browser and are treated as internal navigations.

This fixes bug http://code.google.com/p/chromium/issues/detail?id=52931

Disabled the navigate out chrome frame test as this needs to be fixed to propertly detect
navigations coming out of chrome frame.

Bug=52931
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=57117

[amit, 2010-08-23 16:07:48]

http://codereview.chromium.org/3174027/diff/7001/8001
File chrome/renderer/render_view.cc (right):

http://codereview.chromium.org/3174027/diff/7001/8001#newcode5475
chrome/renderer/render_view.cc:5475: if (!url.ref().empty())
This will keep it in Chrome for _any_ URL with # in it. We want to further
compare if the text before # is the same for new and current URL.

http://codereview.chromium.org/3174027/diff/7001/8001#newcode5480
chrome/renderer/render_view.cc:5480: if (type ==
WebKit::WebNavigationTypeLinkClicked)
is there any link click that should remain in chrome?

[ananta, 2010-08-23 18:02:17]

http://codereview.chromium.org/3174027/diff/7001/8001
File chrome/renderer/render_view.cc (right):

http://codereview.chromium.org/3174027/diff/7001/8001#newcode5475
chrome/renderer/render_view.cc:5475: if (!url.ref().empty())
On 2010/08/23 16:07:48, amit wrote:
> This will keep it in Chrome for _any_ URL with # in it. We want to further
> compare if the text before # is the same for new and current URL.

Removed the anchor check. I verified that anchor navigations don't go through
this codepath.

http://codereview.chromium.org/3174027/diff/7001/8001#newcode5480
chrome/renderer/render_view.cc:5480: if (type ==
WebKit::WebNavigationTypeLinkClicked)
On 2010/08/23 16:07:48, amit wrote:
> is there any link click that should remain in chrome?

Removed this check as well. It appears that the bug this code was intended to
fix has regressed. Will debug that later.

[amit, 2010-08-23 18:21:31]

lgtm. We really need a few good tests to make sure this works as intended.

[Lei Zhang, 2010-08-23 23:38:47]

Did you forget to commit chrome_frame/test/navigation_test.cc ?