Alguns links que comento durante o vídeo:
- Os slides: http://tinyurl.com/aCozinha
Por favor, dêem feedback, pros próximos episódios sairem melhores um pouco ;-)
\o
Desenvolvimento Sustentável de Software



12:12: KentBeck: .@brunopedroso i think you could tdd quicksort. you'd need assertions about the # of comparisons & ways to incrementally improve the count
12:20: the_chrismo: @KentBeck with the karate chop kata (http://bit.ly/aPh2y6) I started to expose comparisons, but felt it revealed internals too much ...
12:23: the_chrismo: @KentBeck ... though without doing those, there's no direct assertion of the efficiency of the algorithm.
12:29: KentBeck: .@the_chrismo i think you'd have to introduce a Comparator & be able to plug in one that keeps a count.
13:55: the_chrismo: @KentBeck ah, nice. makes me want to go back and try that.
12:25: brandheintz: .@KentBeck I'd be interested to see that - would be a crack in the "TDD can't guarantee good algorithms" argument.
12:30 KentBeck :.@bradheintz only people can guarantee good algorithms. tdd can certainly be used to assure that they are used, tho.
13:52: brandheintz: .@KentBeck True, but yr notion of pushing toward better algos through smart use of constraints has appeal.
12:18: davenicolette: @KentBeck @brunopedroso That sounds like it could be a good kata.
12:37: asplake: .@KentBeck @bradheintz TDD for algorithm design seems to verge onto "halting problem" territory
12:38 : KentBeck: .@asplake i don't see why. you need an n log n algorithm. give it n inputs. count the operations. assert that < n log n of them took place.
12:40: asplake: .@KentBeck right. And prove that there are no pathological cases?
12:46 : KentBeck.@asplake that's out of scope for tdd. once you find one, you can write a test to demonstrate it.
14:09: the_real_pablosan: @brunopedroso I have tried growing a binary search from TDD: http://salientblue.com/codenotes/index.html?name=kata2Java01