Guard the rework of documentElement|body.scrollTop/Left under a runtime flag

Guard the rework of documentElement|body.scrollTop/Left under a runtime flag

Blink is now (Nov 11, 2013) at a middle state of cross Web engine compatibility
with regards to documentElement and body scrollLeft/Top properties' behavior [1]:
1. documentElement works as per the spec.
2. body does not, and UseCounts the "improper" usage, keeping the existing
WebKit/Blink quirk behavior.

Given the amount of occurences that rely on the mutually exclusive behavior of
such properties [2], it is preferable to restore the original quirk behavior
and work on the new behavior under a runtime flag: ScrollTopLeftInterop.

In short, as Blink is now, websites that try to work around WebKit/Blink bug by
doing the below math are broken.

scrollOffset = documentElement.scrollTop + body.scrollTop [2]

[1] http://dev.w3.org/csswg/cssom-view/#dom-element-scrolltop
[2] http://bit.ly/17jiCMJ

BUG=314700
R=eseidel

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=162166

[eseidel, 2013-11-11 18:29:57]

It sounds like you want this on for "test" if anything.  Test will make it
enabled for LayoutTests but off for all configurations of chromium.

https://codereview.chromium.org/69143002/diff/1/Source/platform/RuntimeEnable...
File Source/platform/RuntimeEnabledFeatures.in (right):

https://codereview.chromium.org/69143002/diff/1/Source/platform/RuntimeEnable...
Source/platform/RuntimeEnabledFeatures.in:100: ScrollTopLeftInterop
status=experimental
Are you sure you want this on for users with --enable-experimental-web-platform?

[eseidel, 2013-11-11 18:45:07]

I'm OK with it being experimental, or even stable (since that's what it was
before this change).

But you do concern me to tell me that this is half-implemented.

[tonikitoo_, 2013-11-11 19:00:12]

On 2013/11/11 18:45:07, eseidel wrote:
> I'm OK with it being experimental, or even stable (since that's what it was
> before this change).
> 
> But you do concern me to tell me that this is half-implemented.

Right. This is what worries Julien and I too: the half-implementation has
actually made its way to the latest release cut off. We are talking about two
things:

- ToT: if this patch is ok'ed to go to trunk, it addresses the problem there but
introducing the runtime flag, and is a step forward.
- There will be another patch for the stable release branch, where the current
behavior gets reverted to where it was before I start playing with these bits.

[tonikitoo_, 2013-11-11 19:01:57]

> - ToT: if this patch is ok'ed to go to trunk, it addresses the problem there
but
> introducing the runtime flag, and is a step forward.

err.

s/but/by

[eseidel, 2013-11-11 19:02:30]

lgtm

OK.  Lets make sure to fix the branches ASAP.  Do we feel like we have full test
coverage of this change so that we'll knwo when we've fixed the branches
correctly?

[tonikitoo_, 2013-11-11 19:42:12]

On 2013/11/11 19:02:30, eseidel wrote:
> lgtm

Thanks.

> OK.  Lets make sure to fix the branches ASAP.  Do we feel like we have full
test
> coverage of this change so that we'll knwo when we've fixed the branches
> correctly?

We are in good shape with regards to layout test to land this in Blink. What has
to be well coordinated is Chrome's UI code, and  I am handling it for ToT in
https://code.google.com/p/chromium/issues/detail?id=316722 (
https://codereview.chromium.org/68723003/ )

[commit-bot: I haz the power, 2013-11-17 22:50:14]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/a1.gomes@sisa.samsung.com/69143002/110001

[commit-bot: I haz the power, 2013-11-17 23:30:47]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/a1.gomes@sisa.samsung.com/69143002/240001

[commit-bot: I haz the power, 2013-11-18 00:49:30]

Change committed as 162166