Switch from WebDocument::insertStyleText to ::insertUserStyleSheet for programatic CSS injection.

Switch from WebDocument::insertStyleText to ::insertUserStyleSheet for programatic CSS injection.

Inserting CSS as a user stylesheet is both consistent with content script
injection and better for compatibility, since it does not affect the page's DOM.

Removes the other call of insertStyleText (for infobar CSS injection),
thus it should be possible to remove it after a WebKit roll.

Depends on http://webkit.org/b/65158

BUG=82220
TEST=no
R=mpcomplete@chromium.org


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=94499

[Matt Perry, 2011-07-26 23:07:29]

LGTM

http://codereview.chromium.org/7517001/diff/1/chrome/test/data/extensions/api...
File chrome/test/data/extensions/api_test/executescript/basic/test.html (right):

http://codereview.chromium.org/7517001/diff/1/chrome/test/data/extensions/api...
chrome/test/data/extensions/api_test/executescript/basic/test.html:81:
chrome.tabs.get(tabId, pass(function(tab) {
I think this test could end up being flaky, unfortunately. executeScript fires
its callback as soon as the script is sent to the renderer - it does not wait
for the script to actually run. So tabs.get() could end up querying the title
before the script has a chance to update it.

Then again... it looks like the rest of the tests in this file do the same
thing, so maybe I'm mistaken. Just an FYI.

[Mihai Parparita -not on Chrome, 2011-07-26 23:18:01]

On Tue, Jul 26, 2011 at 4:07 PM,  <mpcomplete@chromium.org> wrote:
> I think this test could end up being flaky, unfortunately. executeScript
> fires its callback as soon as the script is sent to the renderer - it
> does not wait for the script to actually run. So tabs.get() could end up
> querying the title before the script has a chance to update it.

As far as as I can tell, ExecuteCodeInTabFunction waits until it gets
a ExtensionHostMsg_ExecuteCodeFinished IPC from the renderer (which is
sent after WebView::executeScriptInIsolatedWorld runs) before sending
the response, so I think this is OK.

Mihai

[commit-bot: I haz the power, 2011-07-28 03:55:31]

Presubmit check for 7517001-4001 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Missing LGTM from an OWNER for:
content/renderer/render_view.cc,content/renderer/render_view.h,content/common/view_messages.h

[Mihai Parparita -not on Chrome, 2011-07-28 03:59:23]

John, could you approve?

Thanks,
Mihai

On 2011/07/28 03:55:31, I haz the power (commit-bot) wrote:
> Presubmit check for 7517001-4001 failed and returned exit status 1.
> 
> Running presubmit commit checks ...
> 
> ** Presubmit ERRORS **
> Missing LGTM from an OWNER for:
>
content/renderer/render_view.cc,content/renderer/render_view.h,content/common/view_messages.h

[jam, 2011-07-28 16:49:23]

content lgtm

[commit-bot: I haz the power, 2011-07-28 18:04:06]

Change committed as 94499