Lecture 29: The Trouble with Concrete Types - Richard Buckland UNSW

The first 17.5 minutes are a discussion of what the task2 diaries revealed: poor time management (eek!) Richard confesses he is bad at time management too and makes some suggestions. The remainder of the lecture is setting up for ADTs (introduced in the next lecture). Task2 (the sudoku solver) is used as motivation. Why do we want to break the problem into separate quasi-independent files? (A: Metcalf's law) What was the relation between the sudokoGrid type and its interface? What is the subtle problem Richard keeps alluding to with respect to the way this separation was implemented? Another Richard, or perhaps Alex, comes up with a better way of implementing sudukoGrid.c - is this as wonderful as it seems or have we unearthed a mare's nest? If only there was some way of solving this problem ... (dissolve to lecture 30)