My biggest problem is knowing whether an idea is worth spending time or even possible. Or just parts of it. There is no place to throw in questions like:

"I need this in my database but im pretty sure it won't scale, will it?

"I wrote this small gadget to solve a very specific problem that I use quite often, does anyone else have that problem?"

"I'm using library X, how would I subclass Y to allow path prefixes?"