Fix graphics corruption that can occur in the PDF viewer when scrolling

Fix graphics corruption that can occur in the PDF viewer when scrolling

In a PDF, if we scroll in a reverse direction to the direction we originally scrolled
(within the same paint) and there were invalidations that happened in-between we
may end up incorrectly clipping the invalidated rects (see crbug.com/488390). This bug
doesn't exist in the original implementation
(ppapi/utility/graphics/paint_aggregator.cc) which uses a different method
of handling invalidations that occur after a scroll. The problem is that
when we scroll the invalidated region, we clip it to the scroll rect. This
can cause us to lose information about what the invalidated region was if
it gets scrolled back into view. We either need to not do this clipping or
disallow combining scrolls that occur in different directions with
invalidations that happen in-between.

BUG=488390
Committed: https://crrev.com/60dbedb323463b79e3c254c38e716efc2d31fbbe
Cr-Commit-Position: refs/heads/master@{#332354}

[raymes, 2015-05-28 05:51:32]

raymes@chromium.org changed reviewers:
+ sammc@chromium.org

[raymes, 2015-06-02 03:01:29]

Hey Sam, PTAL :)

[Sam McNally, 2015-06-02 03:29:26]

https://codereview.chromium.org/1160513004/diff/20001/pdf/paint_aggregator.cc
File pdf/paint_aggregator.cc (right):

https://codereview.chromium.org/1160513004/diff/20001/pdf/paint_aggregator.cc...
pdf/paint_aggregator.cc:13: // We don't use std::sign here to avoid any issues
with negative 0.
Don't pp:Points contain ints?

https://codereview.chromium.org/1160513004/diff/20001/pdf/paint_aggregator.cc...
pdf/paint_aggregator.cc:14: int32_t sign(int32_t num) {
How about
bool IsNegative(int32_t num) { return num < 0; }
(or something similar) instead?

[raymes, 2015-06-02 06:57:08]

https://codereview.chromium.org/1160513004/diff/20001/pdf/paint_aggregator.cc
File pdf/paint_aggregator.cc (right):

https://codereview.chromium.org/1160513004/diff/20001/pdf/paint_aggregator.cc...
pdf/paint_aggregator.cc:14: int32_t sign(int32_t num) {
On 2015/06/02 03:29:26, Sam McNally wrote:
> How about
> bool IsNegative(int32_t num) { return num < 0; }
> (or something similar) instead?

Done.

[Sam McNally, 2015-06-02 06:57:57]

LGTM

[raymes, 2015-06-02 06:58:13]

The CQ bit was checked by raymes@chromium.org

[commit-bot: I haz the power, 2015-06-02 06:58:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1160513004/40001

[commit-bot: I haz the power, 2015-06-02 07:57:22]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-06-02 07:58:13]

Patchset 3 (id:??) landed as
https://crrev.com/60dbedb323463b79e3c254c38e716efc2d31fbbe
Cr-Commit-Position: refs/heads/master@{#332354}