Slick Forums

Output: false


Besides that, another point [9, 0] is not contained either!

This causes problems for example in tile maps.
You have point near to the division line of two tiles and no tile contains it.

I have tried same example with Java2D and it behaves better I think.

Output: true

Java 2D can handle not only [9, 0] but even floats like 9.3[/code]

This is causing more troubles when testing a rectangle on intersection with line that goes along (near to) rectangle's border.

_________________
Jattra

It's the point 9,0 really "contained", I mean it's an edge

Not sure. Probably wrong.

Kev

There could be a philosophical discussion about it.
However, from technical point of view, it is contained. Or do You want to create games in an evnironment, where none of your map tiles contains a point definitely laying on Your map?

_________________
Jattra

In my opinion, given whole number coordinates in pixels. If you were to fill a rectangle, every pixel location that is drawn should be considered to be contained in the rectangle.

Joe

Slick tries to adhere to Java2D in a way (if I remember Kev's original goals correct).

Here's the definition from the 1.4 (Slick's currently supported JDK v) Java2D API:

(Notice how they explicitly define this in the Javadoc)


http://java.sun.com/j2se/1.4.2/docs/api ... 0double%29

So, according to this, point (0,0) should be contained within a rectangle (0,0,10,10), but point (9,9) not!

I don't agree with You appel in one thing. The point (9,9) should be contained as well (my opinion). The key to this is to realize, what the boundary is. Don't You think that each rectangle has one pixel border all around. Don't You think that between two neighboring rectangles is two pixels wide area of nobody, do You? OK, You are suggesting to make one of them belonging to its rectangle. But what about the other one?

So, as I understand it, there are no border pixels, there are just virtual borders. The boundary spoken in the Javadoc is meant to be the pixel with index equal to the rectangle size.

Imagine three appended rectangles of size 10. Let's get it just in X direction to keep it clean.
The bounds are x=0, x=10, x=20. So the 0 belongs to the first one, 10 belongs to the second and make a guess which rectangle should contain the 20. In this scenario, 9 is definitely part of the first one.
That is my view of things.

If Slicks is going along with Java2D, try the example code from my first post. Thank You guys for will to discuss these things.

_________________
Jattra

Point(9,9) cannot belong both to Rectangle(0,0,10,10) and Rectangle(9,9,10,10).

Imagine a grid consisting of rectangles, where no rectangle overlaps another. You have to detect on which rectangle your mouse was over. Can Point(9,9) (your mouse) be contained in TWO rectangles?

appel,

If you were to draw these two rectangles:
first, draw Rectangle(0,0,10,10)
next, draw Rectangle(9,9,10,10)

The second rect would draw over the lower-right corner of the first rect.

In other words, the point (9,9) is contained in both rectangles. Your two rectangles are not adjacent, they are overlapping.

This does not seem to be a bug. As I understand it, it is functioning as designed, but I think it bears looking into. To me, in a graphics lib, it makes sense to treat point containment the same as rendering. If the pixel would be drawn by a filled rectangle, the point is contained within the rectangle.

Joe

No You can't. Neither I can. Neither anybody... oh
Hey, one thing Your two rectangles ARE over-leaping. Count with me. First rectangle starts with zero and is ten points wide OK?
0 1 2 3 4 5 6 7 8 9

And the second starts with nine, so
9 10 11 12 13 14 15 16 17 18

Yes point 9 really belongs to both because they are over-leaping.

_________________
Jattra

Careful with the attitude.

Kev

If You've got me bad, take my apologies. It was not intended. Just trying to make it clear.

One polite question. Have You tried already how Java2D is behaving in thees things? Appel has been referring to it.

_________________
Jattra

In math, I believe, this is symbolized as:

(FILLED O1)--------(EMPTY O1)(FILLED O2)--------(EMPTY O2)

O1 and O2 do not overlap.


Again. The Java2D crowd at Sun bothered putting in text explaining this in a explicit manner. Those people are a lot smarter than you or me, and they must have valid reasons for doing this.

You haven't tried appel, have You? Look It is not so hard. Just copy paste and whole thing is clear.

_________________
Jattra

Interesting!

I need some food to power my brain now

Hello there again.
I am awaiting the final trial. Will this issue be considered to be a bug or not?

We have discussed here and (I hope) finally agreed that Slick is not behaving in accordance with Java2D and in my opinion it is not behaving correctly.

Slick Rectangle(0, 0, 10, 10) draws pixels 0 to 9 but does not contain (calling Rectangle.contains(point) ) neither 0 nor 9.

Last note: I have tried that with Slick distributed as JAR package. I don't have checked latest sources out. Can somebody confirm this behaviour using latest slick? If not, I will do it tonight.

EDIT: Just tried it with latest slick from SVN. It behaves right as JAR distributed version.

_________________
Jattra