Steve Jones - SSC Editor (10/7/2013)
simon.crick (10/7/2013)
Cadavre (10/7/2013)
simon.crick (10/7/2013)
I will probably get shot down again for saying this, but I believe people should stick to core technology and basic tecnhiques unless there is a very good reason for using exotic technology and advanced techniques.Why? Because no one person (even those of you who think you are expert experts) can ever know more than a tiny fraction of all the available technology and techniques, and if we all go off on our own paths following our own preferences for exotic technology and advanced techniques, then our systems will end up being a completely unmaintanable mish-mash of technologies and techniques.
I am not dumb. I have consistently achieved very high grades and won awards for outstanding achievement, etc., but the more I learn, the more I realize it is impossible to know everything, and therefore, out of consideration for our colleagues, we really ought to be sticking to core technology and basic techniques wherever possible.
Simon
I disagree completely. You go with the "best" technique (measurable by performance tests) for the job, regardless of complexity. Then make sure that you document how the logic works.
So if it works 1% faster but is 10 times as complex, you still go for the "advanced" technique?
I'm with Simon here. You stick with what works, what is simplist (KISS principle) and what you can get done quickly. You look for better ways to do things, but you also don't just implement something just because it's faster. Mostly because faster isn't always true. We don't operate on a linear scale, but rather a multi-dimensional one. The stuff that we do might work well for some queries/situations, and not others. It works at some data volumes, but not others.
You look for better techniques and learn how to use them, but you carefully change what you know works well and is simple only when you have good reasons.
Simple techniques does not imply shoddy techniques. Basics are not wrong, they are just of less complexity.
I agree with Simon, as well, but certainly not for the reason you stated, Steve. There is absolutely no reason why "core" or "simple" ever need be exclusive of "fast" and "efficient", which is second only to accuracy and nearly at the same level, BTW (if you don't think so, look at all the "my code is running slow" post on your own forum!). I absolutely agree that a 1% gain in performance can almost never be justified by 10x complexity but that's not the normal case. The normal case is 10x complexity and 60x worse performance. And THAT's the real problem. A lot of people just don't spend the time to learn the core, simple methods which are usually also the very fastest methods.
--Jeff Moden
Change is inevitable... Change for the better is not.