I've always done well following Don Lancaster's advice: Hit the basics, HARD.
Most modern software "innovations" are really just a restatement of David Parnas' paper from 1971. He suggested programs should be split up so that the data used by modules can't be affected by the other modules. "Data hiding."
Once you get a grounding in the history and literature of programming (which all other professions do without thinking), you'll start to see the "sameness" among all the supposed difference. Once you see the "sameness", you'll know what is truly important to know/master, and what is not.
As a recent grad, I'll tell you that the first thing I learned was to ignore all the "software engineering" stuff that schools teach. The reality is that I would be asked, "Build a tool to do X". So I'd start through the process, building a design and the data flows, starting to build the code. Then I'd get "It also has to do Y." And I'd have to throw away most of what was written.
Eventually, I learned the two rules Ken Thompson taught:
1) give users something useful in two week, not the whole problem solved. 2) when in doubt, use brute force.
That's it. With attention to fundamentals, you can clearly see that "Agile" is just a bunch of fancy words (and vendors selling tools) telling you to do that same thing. Only Ken talked about that in the 70's.
What you'll find is that getting a tool into the hands of users will generate new features based on what it already does. No more blue-sky wishes. And it will allow the tool to evolve into what they really want it to do. Which no one ever knows when they start. All the best software followed that path to greatness. (Which you'll see when you read the history of your chosen profession.)
So learn the history. Learn how computers work (not the "Von Neumann" stuff). Get a computer engineering book and learn about bit-slicing and clocks and microcode. It doesn't matter that you're not a hardware guy; you need to know what your code is actually doing, down on the metal.
Never stop learning. Read constantly. Read books you'll never find a job with, like SICP. (I once had to write a loop in a language without a loop construct. But I had read SICP, so I knew if it supported recursion, I could make a loop. And I did.) Don't read trash. Don't bother with the "current hotness"; if it's good, it's probably just a restatement of the old and if it's bad, it'll go away.
I'll just leave with an ancient aphorism: 40 can't tell 20. There's just a gap between what an older person has learned about the world, and what a younger person imagines the world to be. That different can be called "experience", but the fact is I remember perfectly the pearls of wisdom dropped on my by the "old men"...and 20 years later, I was them. So I offer this small essay, knowing it'll probably be rejected, but hoping that just this once it does some good for someone.