Post #2430690

@dws@infosec.exchange @davidism@mas.to Yeah this cleanly sums up my take on it as well. I've found that it's often useful to distinguish between some piece of data being unknown, and being known to be empty/unset/invalid/absent/not applicable/etc. And it can definitely happen that you set up your data models and work on your app for six months before coming across a situation where that distinction is useful. Or, you might not have any way to make the distinction at first, but after a year of development you change your input pipeline in some way that now allows you to distinguish between empty and unknown. So I would definitely plan for it from the beginning, assuming you don't incur any great hardship by doing so.