The Crunch Mode Paradox: Turning Superstars Average
jamesgolick.com
jamesgolick.com
I ended up not doing things the right way, and merely plowed ahead, until I got it to work. However, after the relief of finally getting it done, I have to go back and spend some time refactoring and cleaning up the mess that I made. I often have to remind myself that good engineering simply takes time, and will save in the long run.
But usually I'm always thinking of the structure of the code as I code, continuing the work I did with modeling (if I did any) as I think through what is important, what goes where, what are the important nouns and verbs of my intermediate language. I drop back out of design mode to pick up various pieces of functionality, then continue the work. I'm usually looking for the biggest headache first -- might as well suffer now rather than later. Usually the biggest headaches are the ones that have greatest impact on whether your strategies are correct or not -- whether caching that huge list is going to work real-time or whether a distributed linked-list is going to work. But then I always end up spending the same amount of time on such vexing stuff as "How did I get that box blue, anyway?" or "why is that off by one pixel?" or something like that. Gee, I love programming.