I’ve been growing software program for a few years now. A long time. I’ve realized quite a bit alongside the way in which, and I dare to suppose that I’ve collected some knowledge about software program growth price sharing with others. In some unspecified time in the future, you suppose “Yeah, I’ve bought a grip on this factor.”
And each from time to time, you’ve some revelation that’s the end result of what you’ve realized alongside the way in which. I’ll always remember the second I understood dependency injection. Of course you’ll need to move in an abstraction somewhat than create an implementation!
I had the same revelation this week. I’m certain there are many sager builders amongst you who’ve already figured this out, but it surely struck me immediately that basically good software program design is all about conserving your choices open for so long as potential—that it is best to defer any choices to the final minute.
Now, this appears counter-intuitive. We’ve all the time been instructed to plan the whole lot out forward of time. That we shouldn’t begin the challenge with out having thought of each element that may be thought of. That we have to determine it out now so there aren’t any surprises.
I can simply think about the dialog:
Engineering director: “How’s the planning occurring the brand new challenge?”
Improvement lead: “Nice, we’re able to get began.”
Engineering director: “What database did you choose?”
Improvement lead: “Haven’t determined but.”
Engineering director: “Authentication?”
Improvement lead: “Not but.”
Engineering director: “Wait… what?!”
