tools/source/generic/poly.cxx | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
New commits:
commit 6419ecb40a025f1b4f7df5d059bfca3579dab4d7
Author: Mike Kaganski <mike.kagan...@collabora.com>
AuthorDate: Wed Sep 15 12:34:12 2021 +0200
Commit: Mike Kaganski <mike.kagan...@collabora.com>
CommitDate: Wed Sep 15 16:54:19 2021 +0200
No need to check if nDX is 0 for atan2
We rely on IEEE floating-point arithmetic anyway, and then even both
arguments equal to 0 do not result in an error.
Even if it was different, it would be better to check id both nDX and
nDY are 0, and return 0 early.
It was so ever since "MWS_SRX644: migrate branch mws_srx644 -> HEAD"
commit fd2cf3dc7cd9c73070fa4d70c8ca99c9fc1ce135.
Change-Id: I5b8e2a359374dd1500b149d74eba4c0b0e5cd8d8
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/122115
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kagan...@collabora.com>
diff --git a/tools/source/generic/poly.cxx b/tools/source/generic/poly.cxx
index 70bafc824404..ab33eb61021b 100644
--- a/tools/source/generic/poly.cxx
+++ b/tools/source/generic/poly.cxx
@@ -58,7 +58,7 @@ static double ImplGetParameter( const Point& rCenter, const
Point& rPt, double f
{
const double nDX = static_cast<double>(rPt.X()) - rCenter.X();
const double nDY = static_cast<double>(rCenter.Y()) - rPt.Y();
- double fAngle = atan2(nDY, (nDX == 0) ? 0.000000001 : nDX);
+ double fAngle = atan2(nDY, nDX);
return atan2(fWR*sin(fAngle), fHR*cos(fAngle));
}
