Scaling and Overlap not triggering?

[Sumo]

So I am starting to make a randomly generated dungeon engine. I am creating rooms of random size (with a min/max width and height) that are randomly positioned around the frame. The room object is a 1x1 pixel and I'm using the x and y scale to scale it to the random sizes.

The problem is when I create a room on top of another room, I want to delete it and try again. For some reason the overlap condition doesnt seem to be triggering. I am trying to compare two of the same object for overlap. I'll post a picture of my code and the test frame when I run the program. It's only 3 lines of code so far so something is up...

My Code:


Test Frame:

[zip2kx]

Try doing a seperate, nested loop to check for overlap.

[danworth]

On line 3 try switching the "Index of" and "Loop Index".

[schrodinger]

In line 3 is it object overlaps different instance of same object?
I think that won't work without a separate check as zip2kx suggests
(will also need an extra detector object to catch the overlap)

...but won't also work correctly unless you add a frameskip due to the "overlap delay" bug
(but you can't skip frames unless you change your engine and make it "slower" in building the maze, by allowing 1 frame redraw for each room you place )

If those are strictly "rectangular" rooms you could consider doing "math checks",
won't need extra detectors, will be performance friendly, and won't suffer from the overlap bug

[danworth]

I mean changing it to "Loop Index..." = "Index of..."

[Sumo]

So the problem is having the same instances of an object trying to check against itself? If I create a detector object and have it scale to the newly created room size and location, then test that detector for the previously created rooms, would that change anything? If it returns false, I could then create the normal room in the same location.

What is this "overlap delay" bug? I've never heard of it. I'll try the other suggestion you guys made when I get home from work.

[schrodinger]

If I create a detector object and have it scale to the newly created room size and location, then test that detector for the previously created rooms, would that change anything?

this should theorically work, but won't work due to the overlap delay bug
this bug prevents a correct overlap detection among two "moving" (or scaled, rotated) objects
because the collision mask of the second object ("B" in a "A overlaps B" condition)
will not be refreshed until the next frame cycle

say you have object A at 0,0 and object B at 100,100

then you do:
>>> place B at 0,0

test for A overlaps B
>>> do something

it won't return true until next frame cycle

if A was there since previous frame cycle,
you move B and test B overlaps A
this would return correct because A collision mask is still lagging
but A didn't move so it's still correct

watching your code again, this could possibly work due to the fact you are creating objects on the fly
(created objects forcedly have a "refreshed" collision mask - the bug won't show if you check A against a "just created" B)
but not sure due to the fact you are scaling them, and I think this could cause the issue!