[ollysb]

Half the battle of writing software is crystallising what it should do, in detail. As you discover new edge cases and assimilate them into your code it's very easy to cling on to your initial vision for the architecture. When you return later on and look at the finished product it's all too easy to see that there's a better way to do things. Often you already know this by the time you're finished but you hit a deadline and the next feature/project takes priority over the refactoring.