monline

I've now read quite a few forum topics on the subject and most people seem to argue for one or the other method (the "+" operator, StringBuffer, StringBuilder or whatever). Only a very few posts reflected my opinion: you use the method that makes your code the most robust and easiest to maintain. Then you create a dataset that resembles real-life application and run some tests. If the code runs outside the expected time limits, then you drive it through a profiler, find what makes it crawl and fix it. There's not much purpose in writing the "ultimate" code (ie. spending a lot of time on overall optimization) if only 10% will be used in 90% of the time. And don't take me wrong: there're good coding practises that you should learn and follow. But as with everything else: they are worth only as much as they help you get home earlier. The point of all coding is to create software that fits the requirements of the customer/user/etc. If you write code that is superior to that in any way, then it's either just a waste of time or it's only good for its "inner beauty". Of course the latter can be satisfying too for a developer from time to time ...
I pretty much like what Jeff Atwood had to say on the topic in his blog. And I'm most grateful that somebody else took the time to prove me right.