[WIP] Move layout states into the DocumentLifecycle state machine

[WIP] Move layout states into the DocumentLifecycle state machine

BUG=

[abarth-chromium, 2014-02-06 03:28:34]

Currently ASSERTs in fast/dom/partial-layout-block.html

[ojan, 2014-02-06 05:11:39]

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/Document.cpp
File Source/core/dom/Document.cpp (left):

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/Document.cpp...
Source/core/dom/Document.cpp:468: , m_overMinimumLayoutThreshold(false)
Should probably remove this in a separate pre-patch.

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
File Source/core/dom/DocumentLifecycle.cpp (right):

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
Source/core/dom/DocumentLifecycle.cpp:97: if (state == InStyleRecalc)
Do we really go from LayoutClean to InStyleRecalc without a pitstop in
StyleRecalcPending?

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
Source/core/dom/DocumentLifecycle.cpp:110: return state == InPreLayout;
This is for partialLayout?

https://codereview.chromium.org/156413002/diff/1/Source/core/frame/FrameView.cpp
File Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/156413002/diff/1/Source/core/frame/FrameView....
Source/core/frame/FrameView.cpp:815:
lifecycle().advanceTo(DocumentLifecycle::StyleClean);
I wonder if updateStyleIfNeeded should be doing this. Not clear.

[abarth-chromium, 2014-02-06 05:31:15]

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/Document.cpp
File Source/core/dom/Document.cpp (left):

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/Document.cpp...
Source/core/dom/Document.cpp:468: , m_overMinimumLayoutThreshold(false)
On 2014/02/06 05:11:39, ojan wrote:
> Should probably remove this in a separate pre-patch.

Yep!

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
File Source/core/dom/DocumentLifecycle.cpp (right):

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
Source/core/dom/DocumentLifecycle.cpp:97: if (state == InStyleRecalc)
On 2014/02/06 05:11:39, ojan wrote:
> Do we really go from LayoutClean to InStyleRecalc without a pitstop in
> StyleRecalcPending? 

I'm not sure.  We can split out those two states to see what transitions they
have.

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
Source/core/dom/DocumentLifecycle.cpp:110: return state == InPreLayout;
On 2014/02/06 05:11:39, ojan wrote:
> This is for partialLayout?

This case comes up for scroll bars.  After layout, we resize the scrollbars,
which causes us to need to layout again.  When we unwind the stack, everything
is clean.

https://codereview.chromium.org/156413002/diff/1/Source/core/frame/FrameView.cpp
File Source/core/frame/FrameView.cpp (right):

https://codereview.chromium.org/156413002/diff/1/Source/core/frame/FrameView....
Source/core/frame/FrameView.cpp:815:
lifecycle().advanceTo(DocumentLifecycle::StyleClean);
On 2014/02/06 05:11:39, ojan wrote:
> I wonder if updateStyleIfNeeded should be doing this. Not clear.

Yeah, I wondered that too.  I tried a few variations on that idea, but this
seemed simpler for the time being.  Some of the tricky cases involve calling
updateStyleIfNeeded while the document is inactive.  In those cases, we don't
need to update the style but we also don't end up in the StyleClean state.

[ojan, 2014-02-06 20:36:09]

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
File Source/core/dom/DocumentLifecycle.cpp (right):

https://codereview.chromium.org/156413002/diff/1/Source/core/dom/DocumentLife...
Source/core/dom/DocumentLifecycle.cpp:97: if (state == InStyleRecalc)
On 2014/02/06 05:31:15, abarth wrote:
> On 2014/02/06 05:11:39, ojan wrote:
> > Do we really go from LayoutClean to InStyleRecalc without a pitstop in
> > StyleRecalcPending? 
> 
> I'm not sure.  We can split out those two states to see what transitions they
> have.

Yeah, that was what I had in mind. It's was a bit surprising to me to group
these two states together.

[eseidel, 2014-02-06 23:41:50]

This is amazing.  If this passes tests, lgtm.

[abarth-chromium, 2014-02-06 23:47:44]

On 2014/02/06 23:41:50, eseidel wrote:
> This is amazing.  If this passes tests, lgtm.

Thanks!  I'm breaking it down into smaller pieces.

[abarth-chromium, 2014-02-07 01:55:45]

I'm closing this CL in favor of https://codereview.chromium.org/146023008/

There's a tiny bit more in this CL, but I'll fold that into some other change.