The height-for-row method is used to interrogate the delegate about cell’s height BEFORE the cells are created. This means that, if we are going to have dynamic height cells for any reason whatsoever, some sort of ‘shell’ of the performLayout method we’d be calling from within the cell itself is going to be necessary.

This quickly becomes a maintenance nightmare! Tuning cell layouts pixel-by-pixel means a minimum of two edits to ensure that these two completely different methods always return the same values. These numbers can’t be pulled directly from cells either, unless we create a cell inside this method and have it perform its layout, which is a terribly expensive operation as it will create a handful of objects and size them and calculate line sizes and so on.

There is a possible middle-ground, that will be able to give us cell heights as cheaply as possible and still allow the method to be implemented once . . . but I’ll write up my own article for that soon. ;3

Even using a lazy loading method to computer these #’s on the first pass would still lead to chunky scrolling until the entire table has been scrolled through.