Qt-interest Archive, December 2006
[QT4.2.2] QLineF intersect() request
Message 1 in thread
Is it possible to add a method to QLineF to pass intersect() a
QPointF as well?
Jeff
--
[ signature omitted ]
Message 2 in thread
Hi,
If you mean "add a method in a future release of the library itself", then
you should post your request on the Task Tracker (see the website).
If you mean "add a method in a subclass of my own", I'd say it is not
relevant since QLineF is most often handled statically, not through a
pointer.
Anyhow, to me, the method should perhaps be called "contains" instead of
"intersect" : a line does not "intersect" a point, to me.
You may implement your own method, eventually static. Mine looks like as far
as i remember :
bool contains(QLineF line, QpointF point) {
QRectF rectLine(line.p1, line.p2);
if( ! rectLine.contains(point) )
return false
QLineF line2(point, line.p1)
return (angle(line, line2) == 0);
}
You may have to check whether or not the angle between a null line and a
line is 0 or not, since point may be equal to line.p1 or line.p2
Best-
Nicolas
Jeff Lacki wrote:
>
> Is it possible to add a method to QLineF to pass intersect() a
> QPointF as well?
>
> Jeff
>
> --
> To unsubscribe - send a mail to qt-interest-request@xxxxxxxxxxxxx with
> "unsubscribe" in the subject or the body. List archive and information:
> http://lists.trolltech.com/qt-interest/
--
[ signature omitted ]
Message 3 in thread
Nicolas Castagne wrote:
> You may implement your own method, eventually static. Mine
> looks like as far
> as i remember :
> bool contains(QLineF line, QpointF point) {
> QRectF rectLine(line.p1, line.p2);
> if( ! rectLine.contains(point) )
> return false
>
> QLineF line2(point, line.p1)
> return (angle(line, line2) == 0);
> }
I am not sure I can beat that in terms of inefficiency.
I am tempted to try nevertheless [completley untested]:
bool contains(const QLineF & l, const QPointF & p)
{
const double x0 = p.x();
const double y0 = p.y();
const double x1 = l.p1().x();
const double y1 = l.p1().y();
const double x2 = l.p2().x();
const double y2 = l.p2().y();
const double d = x0 * y1 - y0 * x1
+ x1 * y2 - y1 * x2
- x0 * y2 + y0 * x2;
if (d != 0) // FIXME
return false;
if (x1 < x2)
return x1 <= x0 && x0 <= x2;
if (x2 < x1)
return x2 <= x0 && x0 <= x1;
if (y1 < y2)
return y1 <= y0 && y0 <= y2;
if (y2 < y1)
return y2 <= y0 && y0 <= y1;
return x0 == x1 && y0 == y1; // FIXME
}
Of course, comparing floating point numbers with '=='
is bound to fail for any non-trivial example.
Regards,
Andre'
--
[ signature omitted ]
Message 4 in thread
Thank you Nicolas and Andre. Ill try this out. I agree contains()
is a much better method name per my request. When I get a chance
Ill try this out and submit a request on trolltech.
Jeff
On Wed, 2006-12-06 at 12:33 +0100, Andrà PÃnitz wrote:
> Nicolas Castagne wrote:
> > You may implement your own method, eventually static. Mine
> > looks like as far
> > as i remember :
> > bool contains(QLineF line, QpointF point) {
> > QRectF rectLine(line.p1, line.p2);
> > if( ! rectLine.contains(point) )
> > return false
> >
> > QLineF line2(point, line.p1)
> > return (angle(line, line2) == 0);
> > }
>
> I am not sure I can beat that in terms of inefficiency.
> I am tempted to try nevertheless [completley untested]:
>
> bool contains(const QLineF & l, const QPointF & p)
> {
> const double x0 = p.x();
> const double y0 = p.y();
> const double x1 = l.p1().x();
> const double y1 = l.p1().y();
> const double x2 = l.p2().x();
> const double y2 = l.p2().y();
> const double d = x0 * y1 - y0 * x1
> + x1 * y2 - y1 * x2
> - x0 * y2 + y0 * x2;
> if (d != 0) // FIXME
> return false;
> if (x1 < x2)
> return x1 <= x0 && x0 <= x2;
> if (x2 < x1)
> return x2 <= x0 && x0 <= x1;
> if (y1 < y2)
> return y1 <= y0 && y0 <= y2;
> if (y2 < y1)
> return y2 <= y0 && y0 <= y1;
> return x0 == x1 && y0 == y1; // FIXME
> }
>
> Of course, comparing floating point numbers with '=='
> is bound to fail for any non-trivial example.
>
> Regards,
> Andre'
>
> --
> To unsubscribe - send a mail to qt-interest-request@xxxxxxxxxxxxx with "unsubscribe" in the subject or the body.
> List archive and information: http://lists.trolltech.com/qt-interest/
--
[ signature omitted ]
