Hi
I'd like to break into your discussion, so we won't talk about the same in two threads.
Composite shapes should, to outer world, act line normal shapes, which could be moved, resized, used callbacks on and they should act like we expect usual shapes to. Thing is it is not quite clear what is this "expected" behavior and in my opinion it depends on actual usage of composite shapes, especially with resizing and callbacks.
One might want to resize only one border-like shape and keep rest as it is, or to resize all subshapes to new size, or resize all proportionally, like increase width by 50 %.
The same is there for callbacks, when we might want to have callback for every change of extent of any inner shape, or we might want to have callback only when one of them changes extent, or only when whole shape changes extent (because when we change extent of shape in the center of larger one, it does not change extent of whole shape).
I thought about using replacable blocks for composite shapes actions like adding callbacks, resizing or moving composite shape, reaction to moving or resizing sub-shapes etc.
Jan
abergel wrote
But best is to solve the Trachel issue with the callback. I know, but I need more stable environment (I can easily deal with it, but when other people are working with it and are not necessarily devs, I need a more stable environment for them).
Yes!
I have some ideas about having more free composite shape, where I could specify (either with blocks, or having the option to subclass CompositeShape) what kind of behavior I want, but it definitely needs more refining. Because recently I ran into issue that when resizing a composite shape I wanted to resize only two of them (borders) but not another two (icon and label), so I had to abandon CompositeShape for that problem and do it manually.
Ah ah! You are facing some of the trouble I have already experienced with composed shapes. This is indeed a serious beast. Welcome on board!!!
Alexandre
Moose-dev mailing list
Moose-dev@.unibe
-- View this message in context: http://forum.world.st/Roassal-TRCompositeShape-regression-tp4816726p4818430.... Sent from the Moose mailing list archive at Nabble.com.