Add layout tests for universal overflow scrolling

Add layout tests for universal overflow scrolling

This cl adds layout tests that stress the corner cases that the old composited
scrolling path could not handle. Namely, situations where promoting would cause
stacking and clipping errors.

This is blink-side patch 1 of https://docs.google.com/a/chromium.org/document/d/1Ln57_v6_4RmuuvtWnpsfBOUeDFrPsYNFW13dwuQ8UD8.

BUG=254435
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=156725

[hartmanng, 2013-08-08 21:37:34]

jchaffraix and enne, please take a look.

[Julien - ping for review, 2013-08-21 01:34:40]

Very very sorry for the long delay.

Meta-questions: you separated the different tests into separate files but
wouldn't it make sense to test them in one single file? (I usually ask to split
tests but here there is a lot of overhead with having multiple files)

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
File LayoutTests/compositing/overflow/universal-acceleration/fixed.html (right):

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/fixed.html:9:
window.onload = doTest;
This should totally be folded into the JavaScript file as it is repeated a lot
across the files.

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
File
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js
(right):

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js:3:
//   
internals.setUniversalAcceleratedCompositingForOverflowScrollEnabled(true);
This line is probably unneeded :)

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js:9:
var hasGrandchildren = true;
These variables are just magical and make it hard to understand what's going on
(the tests are overriding them).

One way to change that is to wrap the whole code into a function:

function setUpTest(parameters) {
var isFixedPositioned = parameters.isFixedPositioned || false;

}

This would be called inside your HTML and would expose what is needed.

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js:66:
if ((i % 2) == 0) {
We usually don't put curly braces for single line statement.

[Ian Vollick, 2013-08-23 15:02:39]

It took some doing, but I've combined the layout tests. I'd split them up
because I wanted to be able to do pixel tests. Instead, I've shrunk the test
cases and arranged them in a grid so that I can still do pixel comparisons, just
all at once. This required some positioning of fixed pos elements.

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
File LayoutTests/compositing/overflow/universal-acceleration/fixed.html (right):

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/fixed.html:9:
window.onload = doTest;
On 2013/08/21 01:34:40, Julien Chaffraix wrote:
> This should totally be folded into the JavaScript file as it is repeated a lot
> across the files.

Done.

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
File
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js
(right):

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js:3:
//   
internals.setUniversalAcceleratedCompositingForOverflowScrollEnabled(true);
On 2013/08/21 01:34:40, Julien Chaffraix wrote:
> This line is probably unneeded :)
lol. Done.

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js:9:
var hasGrandchildren = true;
On 2013/08/21 01:34:40, Julien Chaffraix wrote:
> These variables are just magical and make it hard to understand what's going
on
> (the tests are overriding them).
> 
> One way to change that is to wrap the whole code into a function:
> 
> function setUpTest(parameters) {
> var isFixedPositioned = parameters.isFixedPositioned || false;
> 
> }
> 
> This would be called inside your HTML and would expose what is needed.

I've rearranged things a bunch and come up with something similar to what you've
suggested that's hopefully a little less magical. We explicitly ask for a
default set of parameters (so it's both clear to readers where to look to find
out what all the parameters are and so that it's easy to add a new parameter
without updating all tests), update it for the particular test and then
explicitly call doTest(parameters). WDYT?

https://codereview.chromium.org/22620002/diff/5001/LayoutTests/compositing/ov...
LayoutTests/compositing/overflow/universal-acceleration/resources/universal-overflow.js:66:
if ((i % 2) == 0) {
On 2013/08/21 01:34:40, Julien Chaffraix wrote:
> We usually don't put curly braces for single line statement.

Done.

[Julien - ping for review, 2013-08-23 18:28:30]

Much better lgtm

https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
File LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html
(right):

https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html:161:
'hasGrandchildren' : true, },
Nit: It would be more readable if the open and close braces were aligned (which
matches what we do for functions).

https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html:211:
'hasGrandchildren' : false, },
There is only 11 configurations which means that we don't test all 16
combinations. It's OK not to do so but it would be nice to mention why (or fix
this!).

https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html:215:
+ "following configurations:\n" ];
Nit: We don't have the 80 characters limit so this could folded into the
previous line.

[Ian Vollick, 2013-08-26 18:07:44]

On 2013/08/23 18:28:30, Julien - OOO until 3 Sept. wrote:
> Much better lgtm
> 
>
https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
> File
LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html
> (right):
> 
>
https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
>
LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html:161:
> 'hasGrandchildren' : true, },
> Nit: It would be more readable if the open and close braces were aligned
(which
> matches what we do for functions).
Done.
> 
>
https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
>
LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html:211:
> 'hasGrandchildren' : false, },
> There is only 11 configurations which means that we don't test all 16
> combinations. It's OK not to do so but it would be nice to mention why (or fix
> this!).
Done.
> 
>
https://codereview.chromium.org/22620002/diff/33001/LayoutTests/compositing/o...
>
LayoutTests/compositing/overflow/universal-accelerated-overflow-scroll.html:215:
> + "following configurations:\n" ];
> Nit: We don't have the 80 characters limit so this could folded into the
> previous line.
Done.

[commit-bot: I haz the power, 2013-08-26 18:08:00]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/vollick@chromium.org/22620002/38001

[commit-bot: I haz the power, 2013-08-26 21:11:26]

Change committed as 156725