Fix off-by-one error in RasterShape of CSS shapes

Fix off-by-one error in RasterShape of CSS shapes

Be consistent about using end-point exclusive intervals. Before this
patch we created an end-point _inclusive_ interval when scanning to
then end of the line, but an end-point _exclusive_ when ending an
interval somewhere along the scan-line because of a triggered alpha
channel threshold.

Examples:

 01234..
[XXXXXXXXXXXXXXXXXXXX] (X means above alpha channel threshold)

would create the interval 0-19. Since we haven't triggered the alpha
channel threshold, and x points at the last index (19) the end point
is included in the interval and we're supposedly creating an end-point
inclusive interval.

But in the other case:

 01234..
[XXXXXXXXXX..........]

would create the interval 0-10, because when we notice that the alpha
channel threshold doesn't hold, we're already outside the interval,
and x is excluded from the interval meaning we supposedly want to
create a end-point exclusive interval.

To make up for the end-point inclusive intervals,
RasterShape::getExcludedIntervals added an increment to get end-point
exclusive intervals, since that is what the caller of
getExcludedIntervals expects. This obviously broke the other case. We
did have a test for the other case, but the off-by-one error was
accounted for in the test.

After this patch, Shape::createRasterShape always creates end-point
exclusive intervals so we can remove the increment and fix the test.

BUG=363126
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=171569

[davve, 2014-04-14 14:36:02]

Fell into this one when working on SVG sizing. Not sure how to  assign reviewers
to the CSS shapes code, but my idea was that Hans could take a look and Eric
possibly rubber-stamp.

[Hans Muller, 2014-04-14 15:29:46]

On 2014/04/14 14:36:02, David Vest wrote:
> Fell into this one when working on SVG sizing. Not sure how to  assign
reviewers
> to the CSS shapes code, but my idea was that Hans could take a look and Eric
> possibly rubber-stamp.

This looks a good change to me.  Could you add a comment to RasterShape.cpp that
points out that the code (is now) generating end-point exclusive intervals?

[philipj_slow, 2014-04-15 11:37:06]

https://codereview.chromium.org/237123002/diff/20001/Source/core/rendering/sh...
File Source/core/rendering/shapes/Shape.cpp (right):

https://codereview.chromium.org/237123002/diff/20001/Source/core/rendering/sh...
Source/core/rendering/shapes/Shape.cpp:205: int endX = alphaAboveThreshold ? x +
1: x;
missing space before :

[davve, 2014-04-15 11:41:27]

https://codereview.chromium.org/237123002/diff/20001/Source/core/rendering/sh...
File Source/core/rendering/shapes/Shape.cpp (right):

https://codereview.chromium.org/237123002/diff/20001/Source/core/rendering/sh...
Source/core/rendering/shapes/Shape.cpp:205: int endX = alphaAboveThreshold ? x +
1: x;
On 2014/04/15 11:37:07, philipj wrote:
> missing space before :

Done.

[philipj_slow, 2014-04-15 11:41:51]

lgtm

[commit-bot: I haz the power, 2014-04-15 11:42:07]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/davve@opera.com/237123002/40001

[commit-bot: I haz the power, 2014-04-15 12:43:44]

Change committed as 171569