It depends what you want to achieve actually. So far, shape dimensions are mapped to some metrics using width: and height:. Mondrian do not cache them. It could, but one would expect that the represented model cache them. If I introduce a cache mechanism, when should I invalidate the cache? This is not clear to me...
Why do you want invalidate the cache ? I think that Mondrian should make a cache when it open to facilitate the navigation, then when the canvas is closed the cache is removed.
If it is done in Mondrian, it will be more compatible with all things that we can do ?
It has been compatible so far. Why would it change?
Yes, what I want to say is that if a slow computation is done in the script, actually Mondrian become slow. If we would have a cache in Mondrian, this problem does not appear anymore.
Cheers --- Jannik