Easier way to obtain object bounds about rs-tiled HOT 5 OPEN

It sounds definitely useful, but should these bounds also take into account object alignment and rotation then?

And I think bounding_rect might be a more fitting name, although the QGraphicsItem::boundingRect function does not include the item's position and rotation.

from rs-tiled.

OK, we could add a local_bounding_rect function that doesn't apply transformations for now, and then we could figure out rotations/position/etc on a global_bounding_rect function. Does that sound good?

from rs-tiled.

I'm not sure about the verbosity of these names. Wouldn't a bounding_rect that just returns the local rectangle be fine, and we could call the global one aabb? But if you prefer the explicit local/global I'm fine with that too.

Btw, you initially proposed only a global_bounds, so I asked whether that should then also take into account alignment and rotation because I thought those things should probably go together. I didn't mean to propose we also add a version returning a local rectangle. Maybe we should wait with adding that until there's an actual need.

from rs-tiled.

Btw, you initially proposed only a global_bounds, so I asked whether that should then also take into account alignment and rotation because I thought those things should probably go together. I didn't mean to propose we also add a version returning a local rectangle. Maybe we should wait with adding that until there's an actual need.

I didn't think of the implications of having global in the name. While it is true that global bounds should include rotation and alignment, it would complicate the implementation. There are times where rotation isn't applied to an object and it is useful to know its bounds, for instance when you want to obtain the width/height of a rect object quickly without having to match like in the example above, so I think having local in the name is best to differentiate between a possible global implementation later down the line.

from rs-tiled.

so I think having local in the name is best to differentiate between a possible global implementation later down the line.

Alright. What about alignment, though? I guess it is relevant even for a local bounding rectangle. For a rectangle with top-left alignment, the local bounding rect would have its top-left at (0,0), but for bottom-left alignment, that rectangle would need to have its top-left at (0,-height), right?

Of course the problem with alignment is that it depends not only on the type of object, but also on the tileset object alignment property as well as on the map orientation.

from rs-tiled.