Fix bug with very large round rects with large radii

Fix bug with very large round rects with large radii

BUG=463920
Committed: https://skia.googlesource.com/skia/+/2a679ae8f5b80a337f67783dbc0a447a9f4312c7

[robertphillips, 2015-03-11 18:12:54]

robertphillips@google.com changed reviewers:
+ reed@google.com

[robertphillips, 2015-03-11 18:13:28]

https://codereview.chromium.org/998903003/diff/20001/src/core/SkRRect.cpp
File src/core/SkRRect.cpp (right):

https://codereview.chromium.org/998903003/diff/20001/src/core/SkRRect.cpp#new...
src/core/SkRRect.cpp:175: // If f < 1, then all corner radii are reduced by
multiplying them by f."
So, if we don't do this math in doubles the code never realizes there is a
problem.

https://codereview.chromium.org/998903003/diff/20001/src/core/SkRRect.cpp#new...
src/core/SkRRect.cpp:588: for (int i = 0; i < 4; ++i) {
This only enforces predicates on individual radii. The failure involves both
radii on a side.

[robertphillips, 2015-03-13 13:51:09]

ping

[reed1, 2015-03-13 14:16:21]

https://codereview.chromium.org/998903003/diff/20001/src/core/SkRRect.cpp
File src/core/SkRRect.cpp (right):

https://codereview.chromium.org/998903003/diff/20001/src/core/SkRRect.cpp#new...
src/core/SkRRect.cpp:178: if (fRadii[0].fX + ((double)fRadii[1].fX) >
rect.width()) {
can we use a helper function for these, one that takes doubles?

[robertphillips, 2015-03-13 16:32:35]

https://codereview.chromium.org/998903003/diff/20001/src/core/SkRRect.cpp
File src/core/SkRRect.cpp (right):

https://codereview.chromium.org/998903003/diff/20001/src/core/SkRRect.cpp#new...
src/core/SkRRect.cpp:178: if (fRadii[0].fX + ((double)fRadii[1].fX) >
rect.width()) {
On 2015/03/13 14:16:21, reed1 wrote:
> can we use a helper function for these, one that takes doubles?

Done.

[reed1, 2015-03-13 16:41:53]

lgtm w/ request for comment explaining the doubles in the helper.

https://codereview.chromium.org/998903003/diff/40001/src/core/SkRRect.cpp
File src/core/SkRRect.cpp (right):

https://codereview.chromium.org/998903003/diff/40001/src/core/SkRRect.cpp#new...
src/core/SkRRect.cpp:131: static double compute_min_scale(double rad1, double
rad2, double limit, double curMin) {
// explicitly use doubles here because ...

[robertphillips, 2015-03-13 16:46:23]

https://codereview.chromium.org/998903003/diff/40001/src/core/SkRRect.cpp
File src/core/SkRRect.cpp (right):

https://codereview.chromium.org/998903003/diff/40001/src/core/SkRRect.cpp#new...
src/core/SkRRect.cpp:131: static double compute_min_scale(double rad1, double
rad2, double limit, double curMin) {
On 2015/03/13 16:41:53, reed1 wrote:
> // explicitly use doubles here because ...

Done.

[robertphillips, 2015-03-13 16:46:29]

The CQ bit was checked by robertphillips@google.com

[robertphillips, 2015-03-13 16:46:30]

The patchset sent to the CQ was uploaded after l-g-t-m from reed@google.com
Link to the patchset: https://codereview.chromium.org/998903003/#ps60001 (title:
"Added comment")

[commit-bot: I haz the power, 2015-03-13 16:46:42]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/998903003/60001

[commit-bot: I haz the power, 2015-03-13 16:53:12]

Committed patchset #4 (id:60001) as
https://skia.googlesource.com/skia/+/2a679ae8f5b80a337f67783dbc0a447a9f4312c7