Details
-
Bug
-
Resolution: Unresolved
-
P3: Somewhat important
-
None
-
5.15.2
-
None
-
All
Description
In the source file qtransform.cpp, the micro Q_NEAR_CLIP is defined a positive tiny number.
68 | #define Q_NEAR_CLIP (sizeof(qreal) == sizeof(double) ? 0.000001 : 0.0001)
Where the Q_NEAR_CLIP is used
1604 | if (ha.w < Q_NEAR_CLIP && hb.w < Q_NEAR_CLIP) 1605 | return false;
will be return false when ha.w and hb.w is negative, when m33 is a negative. So before compare with Q_NEAR_CLIP, number should be round with qAbs().
1604 | if (qAbs(ha.w) < Q_NEAR_CLIP && qAbs(hb.w) < Q_NEAR_CLIP) 1605 | return false;
Here the test code is:
#include <QtGui> int main(int argc, char* argv[]) { QPolygonF a(QRectF(0, 0, 100, 200)); qDebug() << "points count in polygon a is " << a.size(); qDebug() << a; QTransform t1(1, 0, 0, 0, 1, 0, 0, 0, 1); qDebug() << "transform: " << t1; QPolygonF b = t1.map(a); qDebug() << "points count in polygon b is " << b.size(); qDebug() << b; QTransform t2(1, 0, 0, 0, 1, 0, 0, 0, -1); //map will failled when m33 is negative qDebug() << "transform: " << t2; QPolygonF c = t2.map(a); qDebug() << "points count in polygon c is " << c.size(); qDebug() << c; return 0; }
Output as:
points count in polygon a is 5 QPolygonF(QPointF(0,0)QPointF(100,0)QPointF(100,200)QPointF(0,200)QPointF(0,0)) transform: QTransform(type=TxNone, 11=1 12=0 13=0 21=0 22=1 23=0 31=0 32=0 33=1) points count in polygon b is 5 QPolygonF(QPointF(0,0)QPointF(100,0)QPointF(100,200)QPointF(0,200)QPointF(0,0)) transform: QTransform(type=TxProject, 11=1 12=0 13=0 21=0 22=1 23=0 31=0 32=0 33=-1) points count in polygon c is 1 QPolygonF(QPointF(0,0))
We expect the map result c is the same as b, but fail. The bug occur here when the transform.m33 is a negative number.
Attachments
Gerrit Reviews
| For Gerrit Dashboard: QTBUG-118751 | ||||||
|---|---|---|---|---|---|---|
| # | Subject | Branch | Project | Status | CR | V |
| 525734,5 | Handle negative weight m33 parameter in QTransform | dev | qt/qtbase | Status: NEW | -1 | 0 |