diff options
author | Ben Murdoch <benm@google.com> | 2009-08-11 17:01:47 +0100 |
---|---|---|
committer | Ben Murdoch <benm@google.com> | 2009-08-11 18:21:02 +0100 |
commit | 0bf48ef3be53ddaa52bbead65dfd75bf90e7a2b5 (patch) | |
tree | 2943df35f62d885c89d01063cc528dd73b480fea /WebCore/platform/graphics/cg/PatternCG.cpp | |
parent | 7e7a70bfa49a1122b2597a1e6367d89eb4035eca (diff) | |
download | external_webkit-0bf48ef3be53ddaa52bbead65dfd75bf90e7a2b5.zip external_webkit-0bf48ef3be53ddaa52bbead65dfd75bf90e7a2b5.tar.gz external_webkit-0bf48ef3be53ddaa52bbead65dfd75bf90e7a2b5.tar.bz2 |
Merge in WebKit r47029.
Diffstat (limited to 'WebCore/platform/graphics/cg/PatternCG.cpp')
-rw-r--r-- | WebCore/platform/graphics/cg/PatternCG.cpp | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/WebCore/platform/graphics/cg/PatternCG.cpp b/WebCore/platform/graphics/cg/PatternCG.cpp index 697bc57..63628f4 100644 --- a/WebCore/platform/graphics/cg/PatternCG.cpp +++ b/WebCore/platform/graphics/cg/PatternCG.cpp @@ -62,10 +62,10 @@ CGPatternRef Pattern::createPlatformPattern(const TransformationMatrix& userSpac // If FLT_MAX should also be used for xStep or yStep, nothing is rendered. Using fractions of FLT_MAX also // result in nothing being rendered. // INT_MAX is almost correct, but there seems to be some number wrapping occuring making the fill - // pattern is not filled correctly. - // So, just pick a really large number that works. - float xStep = m_repeatX ? tileRect.width() : (100000000.0f); - float yStep = m_repeatY ? tileRect.height() : (100000000.0f); + // pattern is not filled correctly. + // To make error of floating point less than 0.5, we use the half of the number of mantissa of float (1 << 22). + CGFloat xStep = m_repeatX ? tileRect.width() : (1 << 22); + CGFloat yStep = m_repeatY ? tileRect.height() : (1 << 22); // The pattern will release the tile when it's done rendering in patternReleaseCallback tileImage()->ref(); |