On 1/10/17, 1:45 PM, "Christofer Dutz" <christofer.d...@c-ware.de> wrote:
>Hi, > >As we were discussing this in another thread. Wouldn’t layout be one of >the things that our invalidateSize() + measure() logic of the regular SDK >would be beneficial. >Otherwise I bet some layouting operations would be executed serially and >multiple times (sort of executing the logic of “measure” after setting >the width and then again after setting height. IMO, the general purpose practice of setting a backing variable, calling invalidateXXX and waiting for validation for every property was inefficient. Most properties don't need it. But special casing pairs or sets like x,y or w,h or maybe x,y,w,h might be worth it, but I think that also depends. Sometimes you need to set w to determine h and there is no good alternative. Some component set is welcome to put in a general purpose invalidation mechanism, but for the basic set, I'd rather just profile for hot spots and understand what the right optimizations are. My 2 cents, -Alex
