mit 计算机科学及编程导论（11）

words:

fake , ambiguity , readily , physics , theoretical , subtlety , alias , pragmatic , quantum , supreme , validation , nail , unwarranted ,  weave , facilitate , feasible , relay , moth , trap , epiphany , laboratory , diagnosis ,  disabuse , crawl , breed , mate , eliminate , be hung up about/on sth , stamp someone out , whack a mole , tray , cockroach , evolve , probe , subtle , skeptical , refute , flawed , intermediate , slip-up , prune , kneel , ascertain ,  integration 

notes:

two topics : testing(testing & reasoning to check whether the testing is enough or not) & debugging(function & performance)

testing:

1. two ways of testing : unit testing & integration testing 

2. test suit: small enough to test in reasonable time & large enough to give enough confidences that the problem is ok(how to trade off)

debugging :

1. goal : not to eliminate one bug but to move towards a bug-free program (so we need a debug strategy)

2. what we should ask when we debug:

   how could it have produce this result?

   is it part of family?(一类错误)

   how to fix it?

scientific method(我不是很了解，暂时不贴，但是这好像是他们高中教的如何做科学研究的方法)

ps:

how to use a debug strategy to write a bug-free program is very crucial , we must think "is it a family" when we ascertain a bug to eliminate a kind of bug(以前都是一个一个的解决，解决后就运行，不行再找下一个，这样看来还是不够系统systematic)