You're right that the field is moving quickly. Eboni's point about an internship is a great one, because that will make things a lot less abstract.
In general, I would recommend trying to get as firm a foundation as possible. It's true, you don't know what languages companies will be looking for in five years, what platforms will be big deals, what companies will be hiring, and so on.
But you can be pretty certain that it will be valuable to be able to "think procedurally," to be able to break a task down into a series of smaller steps that a computer can follow. It will be useful to know standard algorithms and data structures and their tradeoffs (when should you use a hash table versus a tree? A set versus a list? and so on). The easiest way to get to know these is to try a number of concrete things out, but don't let yourself believe that you've stumbled across the One True Way that will never change.
I hope this helps!