Smalltalk › Pharo › Pharo Smalltalk Users

Rectangle and Polygon classes' bugs?

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

3 messages Options

Andreas Sunardi

Rectangle and Polygon classes' bugs?

Pharo 5.0

It seems like a bug to me, but perhaps I misunderstand something.

1) Comment in Rectangle class >> origin:corner: method shows 'origin' should be top left and 'corner' should be bottom right. This method calls Rectangle >> setPoint:point:. But this method calculates 'origin' to be bottom left corner and 'corner' to be top right. Now all the edge and corner methods are wrong.

r := Rectangle origin: 0@0 corner: 10@10.

r topLeft. => (0@0) "should be 0@10"

r bottomRight. => (10@10) "should be 10@0"

r top. => 0 "should be 10"

r bottom. => 10 " should be 0"

2) Why Polygon >> bounds (which is PathShape >> bounds) returns a rectangle with a corner 1 point larger than the actual boundary?

p1 := Polygon vertices: {0@0 . 0@10 . 10@10 . 10@0}.

p1 bounds. => (0@0) corner: (11@11). "should be 0@0 and 10@10"

This seems to be a fix to Rectangle >> containsPoint: strange behavior. By strange behavior, I mean:

r := Rectangle point: 0@0 point: 10@10.

r containsPoint: 0@0. => true

r containsPoint: 10@10. => false "should be true"

{ r topLeft . r topRight . r bottomLeft . r bottomRight } allSatisfy [ :corner | r containsPoint: corner ]. => false "should be true"

I want to fix this in my image, but Rectangle is used all over the place and I'm worry I'll break a lot of things. I wonder if I just misunderstood how these are supposed to be used. Can anybody help me understand if this is intentional or a bug?

PS: I looked at RectangleTest and I'm not sure I understand why they are testing what they are testing.

Andreas Sunardi

Ben Coman

Re: Rectangle and Polygon classes' bugs?

On Sat, May 20, 2017 at 12:23 AM, Andreas Sunardi <[hidden email]> wrote:

Pharo 5.0

It seems like a bug to me, but perhaps I misunderstand something.

1) Comment in Rectangle class >> origin:corner: method shows 'origin' should be top left and 'corner' should be bottom right. This method calls Rectangle >> setPoint:point:. But this method calculates 'origin' to be bottom left corner and 'corner' to be top right. Now all the edge and corner methods are wrong.

r := Rectangle origin: 0@0 corner: 10@10.
r topLeft. => (0@0) "should be 0@10"
r bottomRight. => (10@10) "should be 10@0"
r top. => 0 "should be 10"
r bottom. => 10 " should be 0"

Would this depend on where you consider 0@0 to be on the screen?

(em := EllipseMorph new) openInWorld.

em bounds "==> (0@0) corner: (50@40)"

2) Why Polygon >> bounds (which is PathShape >> bounds) returns a rectangle with a corner 1 point larger than the actual boundary?

p1 := Polygon vertices: {0@0 . 0@10 . 10@10 . 10@0}.
p1 bounds. => (0@0) corner: (11@11). "should be 0@0 and 10@10"

This seems to be a fix to Rectangle >> containsPoint: strange behavior. By strange behavior, I mean:

r := Rectangle point: 0@0 point: 10@10.
r containsPoint: 0@0. => true
r containsPoint: 10@10. => false "should be true"
{ r topLeft . r topRight . r bottomLeft . r bottomRight } allSatisfy [ :corner | r containsPoint: corner ]. => false "should be true"

I want to fix this in my image, but Rectangle is used all over the place and I'm worry I'll break a lot of things. I wonder if I just misunderstood how these are supposed to be used. Can anybody help me understand if this is intentional or a bug?

Sorry, I don't know about this.

cheers -ben

Andreas Sunardi

Re: Rectangle and Polygon classes' bugs?

Ah, I see, so Rectangle is in the context of display geometry not Cartesian coordinate (I don't know if I'm using the right words here, but I hope you get what I mean). You're right, the package is 'Graphics-Shapes'. For some reason, I thought it said 'Geometry'

I'm building something for layout geometry and these would be the wrong classes to use then. Question #2 would be a moot point.

Thanks for the answer. You've saved me from going to the wrong path.

Andreas Sunardi

On Fri, May 19, 2017 at 9:53 AM, Ben Coman <[hidden email]> wrote:

On Sat, May 20, 2017 at 12:23 AM, Andreas Sunardi <[hidden email]> wrote:
Pharo 5.0

It seems like a bug to me, but perhaps I misunderstand something.

1) Comment in Rectangle class >> origin:corner: method shows 'origin' should be top left and 'corner' should be bottom right. This method calls Rectangle >> setPoint:point:. But this method calculates 'origin' to be bottom left corner and 'corner' to be top right. Now all the edge and corner methods are wrong.

r := Rectangle origin: 0@0 corner: 10@10.
r topLeft. => (0@0) "should be 0@10"
r bottomRight. => (10@10) "should be 10@0"
r top. => 0 "should be 10"
r bottom. => 10 " should be 0"

Would this depend on where you consider 0@0 to be on the screen?
(em := EllipseMorph new) openInWorld.
em bounds "==> (0@0) corner: (50@40)"

2) Why Polygon >> bounds (which is PathShape >> bounds) returns a rectangle with a corner 1 point larger than the actual boundary?

p1 := Polygon vertices: {0@0 . 0@10 . 10@10 . 10@0}.
p1 bounds. => (0@0) corner: (11@11). "should be 0@0 and 10@10"

This seems to be a fix to Rectangle >> containsPoint: strange behavior. By strange behavior, I mean:

r := Rectangle point: 0@0 point: 10@10.
r containsPoint: 0@0. => true
r containsPoint: 10@10. => false "should be true"
{ r topLeft . r topRight . r bottomLeft . r bottomRight } allSatisfy [ :corner | r containsPoint: corner ]. => false "should be true"

I want to fix this in my image, but Rectangle is used all over the place and I'm worry I'll break a lot of things. I wonder if I just misunderstood how these are supposed to be used. Can anybody help me understand if this is intentional or a bug?

Sorry, I don't know about this.
cheers -ben