Fixes and reenables RendererAccessibilityBrowserTest.*

Fixes and reenables RendererAccessibilityBrowserTest.*

Revert "Disable RendererAccessibilityBrowserTest.CrossPlatform* tests"

This reverts commit 2a31da98908e3e42b323f4d67b957fc8b85941e3.

and

ensures that we replace the test tree when receiving layout complete notifications.
Only pass on the test tree when the tree actually has a non-empty subtree. This indicates a non-blank test page.

BUG=106934
TEST=try bots.
TBR=jam,sergyu

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=114046

[dmazzoni, 2011-12-10 06:27:35]

lgtm

http://codereview.chromium.org/8895018/diff/1/content/browser/renderer_host/r...
File content/browser/renderer_host/render_view_host.cc (right):

http://codereview.chromium.org/8895018/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_view_host.cc:1402:
save_accessibility_tree_for_testing_) {
I just realized that there's no reason to go through the loop at all if
save_accessibility_tree_for_testing_ is false, so put this test up next to if
(!params.empty()).

[David Tseng, 2011-12-12 01:23:57]

http://codereview.chromium.org/8895018/diff/1/content/browser/renderer_host/r...
File content/browser/renderer_host/render_view_host.cc (right):

http://codereview.chromium.org/8895018/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_view_host.cc:1402:
save_accessibility_tree_for_testing_) {
On 2011/12/10 06:27:36, Dominic Mazzoni wrote:
> I just realized that there's no reason to go through the loop at all if
> save_accessibility_tree_for_testing_ is false, so put this test up next to if
> (!params.empty()).

I'd like to keep this as is to avoid any possible race conditions. Small changes
in this logic seems to lead to flakyness in the tests, are fairly hard to debug,
and require waiting for try runs to confirm for stability.

With that said, would it make sense to #ifdef this block for only debug builds?

[dmazzoni, 2011-12-12 02:54:15]

On 2011/12/12 01:23:57, David Tseng wrote:
> I'd like to keep this as is to avoid any possible race conditions. Small
changes
> in this logic seems to lead to flakyness in the tests, are fairly hard to
debug,
> and require waiting for try runs to confirm for stability.

That's fine, I'll try it in a separate change.

> With that said, would it make sense to #ifdef this block for only debug
builds?

No, I'd like the tests to be able to run in Release mode.

- Dominic

[commit-bot: I haz the power, 2011-12-12 17:34:31]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dtseng@chromium.org/8895018/1

[commit-bot: I haz the power, 2011-12-12 18:31:12]

Change committed as 114046