Allow Length::nonNanCalculatedValue to take a LayoutUnit

Allow Length::nonNanCalculatedValue to take a LayoutUnit

By allowing Length::nonNanCalculatedValue to take a LayoutUnit 
parameter, we avoid the possible rounding when minimumValueForLength
converted its LayoutUnit to int in the calc() code path.

Behavior matches Firefox.

BUG=427207, 448796, 251627
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=188981

[rwlbuis, 2015-01-20 15:42:36]

rob.buis@samsung.com changed reviewers:
+ jchaffraix@chromium.org, leviw@chromium.org

[rwlbuis, 2015-01-20 15:42:37]

PTAL

[Julien - ping for review, 2015-01-23 10:22:26]

https://codereview.chromium.org/845313010/diff/20001/Source/platform/LengthFu...
File Source/platform/LengthFunctions.cpp (right):

https://codereview.chromium.org/845313010/diff/20001/Source/platform/LengthFu...
Source/platform/LengthFunctions.cpp:74: return
length.nonNanCalculatedValue(maximumValue.toFloat());
How about we just pass a LayoutUnit to the function?

[rwlbuis, 2015-01-23 16:08:49]

On 2015/01/23 10:22:26, Julien Chaffraix - CET wrote:
>
https://codereview.chromium.org/845313010/diff/20001/Source/platform/LengthFu...
> File Source/platform/LengthFunctions.cpp (right):
> 
>
https://codereview.chromium.org/845313010/diff/20001/Source/platform/LengthFu...
> Source/platform/LengthFunctions.cpp:74: return
> length.nonNanCalculatedValue(maximumValue.toFloat());
> How about we just pass a LayoutUnit to the function?

Done! Hopefully it turns out all green.

[Julien - ping for review, 2015-01-26 10:55:22]

lgtm

https://codereview.chromium.org/845313010/diff/40001/LayoutTests/fast/css/cal...
File LayoutTests/fast/css/calc-rounding.html (right):

https://codereview.chromium.org/845313010/diff/40001/LayoutTests/fast/css/cal...
LayoutTests/fast/css/calc-rounding.html:21: </style>
Let's add a description of what this is testing. The condition for passing
should also be added, unless it's converted to a check-layout.js file (see
below).

https://codereview.chromium.org/845313010/diff/40001/LayoutTests/fast/css/cal...
LayoutTests/fast/css/calc-rounding.html:24: <div id="remainderDiv"></div>
AFAICT this could be a check-layout.js test as what this is trying to do is
measure the size of the 2 inner divs. The upside is that this would be clearer
and faster.

[rwlbuis, 2015-01-26 15:40:08]

On 2015/01/26 10:55:22, Julien Chaffraix - CET wrote:
> lgtm
> 
>
https://codereview.chromium.org/845313010/diff/40001/LayoutTests/fast/css/cal...
> File LayoutTests/fast/css/calc-rounding.html (right):
> 
>
https://codereview.chromium.org/845313010/diff/40001/LayoutTests/fast/css/cal...
> LayoutTests/fast/css/calc-rounding.html:21: </style>
> Let's add a description of what this is testing. The condition for passing
> should also be added, unless it's converted to a check-layout.js file (see
> below).

Done.

>
https://codereview.chromium.org/845313010/diff/40001/LayoutTests/fast/css/cal...
> LayoutTests/fast/css/calc-rounding.html:24: <div id="remainderDiv"></div>
> AFAICT this could be a check-layout.js test as what this is trying to do is
> measure the size of the 2 inner divs. The upside is that this would be clearer
> and faster.

Great suggestion, I like check-layout.js, done.

[rwlbuis, 2015-01-26 16:52:59]

The CQ bit was checked by rob.buis@samsung.com

[commit-bot: I haz the power, 2015-01-26 16:53:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/845313010/60001

[commit-bot: I haz the power, 2015-01-26 18:14:39]

Committed patchset #4 (id:60001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=188981