Stop passing NodeRenderingContext except in textRendererIsNeeded

Stop passing NodeRenderingContext except in textRendererIsNeeded

We don't need the NodeRenderingContext in childShouldCreateRenderer or rendererIsNeeded,
we only need the node or the style so pass them explicitly. This makes textRendererIsNeeded
the only place we pass the NodeRenderingContext. Future patches should be able to clean that
up allowing NodeRenderingContext to only be the creator of renderers and not a multi-purpose
bag of things.

This also changes both methods to take a Node& or a RenderStyle& instead of pointers so
users of these APIs know the values can never be null.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=157424

[ojan, 2013-09-08 23:07:57]

lgtm

[commit-bot: I haz the power, 2013-09-08 23:08:04]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/esprehn@chromium.org/23956012/1

[commit-bot: I haz the power, 2013-09-09 00:11:01]

Change committed as 157424

[eseidel, 2013-09-09 03:13:48]

I have mixed feelings about passing RefCounted subclasses by reference.

:shrug:

[esprehn, 2013-09-09 04:45:20]

On 2013/09/09 03:13:48, eseidel wrote:
> I have mixed feelings about passing RefCounted subclasses by reference.
> 
> :shrug:

I'm actually not a fan of it either, but adamk and others had expressed interest
in making it more clear when things couldn't be null by using references.

I'd actually be more interested in adding something like NotNull<T> and then
using that for these kinds of things. Then Node::document() would return
NotNull<Document> which overloads -> and has .get() and the normal RefPtr like
things but doesn't overload operator! so you can't null check it. That would
make things like node->document() == node->document() sensible since you
wouldn't need ampersands on both sides.