It's hard to go wrong with KISS as a basis. The successful design of a module (or library (or interface)) pretty much depends on either experience or empathy or both. This is basically a 'walk a mile in another's shoes' kind of thing and efforts to expand functionality beyond a simple model quickly fractalize into complexity. That said, another area to explore lies below the obvious user level, i.e. below fgrep etc. Sometimes it is not the higher functions a user wants, but rather the primitives used to build the higher level. If these are exposed clearly and cleanly in the module then you appease the user who wants a portion of what you've designed, but has something else in mind in terms of what to use it for. This is surprisingly difficult to do in a generic and style free manner. Which last brings up an important criteria for evaluating the entire module. Namely, how much does the intended user have to think like the module author in order to use the module? Obviously this should be minimized as much as can be without diminishing function.

So, short answer, yup KISS is good, elsewhere there be dragons!

–hsm

"Never try to teach a pig to sing…it wastes your time and it annoys the pig."