Replace check for RenderLayer in RenderLayerModelObject::ensureLayer with ASSERT due to redundant c…

Replace check for RenderLayer in RenderLayerModelObject::ensureLayer with ASSERT due to redundant check.

RenderLayerModelObject::styleDidChange() invokes the ensureLayer() method only in case of !layer().
Again checking for layer existence would be deemed redundant in ensureLayer. Replacing it with an
ASSERT(!m_layer).

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=159776

[vivekg__, 2013-10-15 10:50:43]

Please take a look, thanks!

[vivekg__, 2013-10-16 06:28:01]

On 2013/10/15 10:50:43, vivekg_ wrote:
> Please take a look, thanks!

I am not sure why bots are turning red for this trivial change. My local tests
are working as expected. Even the unit-tests are fine.

[eseidel, 2013-10-16 06:34:49]

Try running the tests in debug. I suspect your assert is failing.  In release,
you'd end up releasing the layer too soon, causing all these odd failures I bet.

[vivekg_samsung, 2013-10-16 07:07:00]

On 2013/10/16 06:34:49, eseidel wrote:
> Try running the tests in debug. I suspect your assert is failing.  In release,
> you'd end up releasing the layer too soon, causing all these odd failures I
bet.

Thank you Eric. Yeah I have tried both Debug and Release and got the same
results. Also ran the unit tests fine.

The reason for this change is on line 152 (http://goo.gl/ZNmdmq), there is a
check for !layer() before calling ensureLayer() on line 155.
So there is no way in which the m_layer exists inside ensureLayer(). Hence I
ended up adding just the ASSERT thereby removing the 
redundant check every time a layer is required.

[eseidel, 2013-10-16 07:08:38]

I see.  There is only one call to ensureLayer():
https://code.google.com/p/chromium/codesearch#search/&q=ensureLayer&sq=packag...

We should consider renaming/refactoring this to createLayer() since it always
creates, but lgtm.

[commit-bot: I haz the power, 2013-10-16 12:25:22]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/vivek.vg@samsung.com/27246003/13001

[commit-bot: I haz the power, 2013-10-16 14:43:52]

Change committed as 159776

[Julien - ping for review, 2013-10-16 14:47:52]

late lgtm

ensureLayer was a remnant of lazily allocating RenderLayer for overflow clip so
it made sense for it actually check if we had a layer. We are not going to
revive this change anyway.