Personally, I think it depends. Its more pragmatic to think about "when is it the right moment to use this approach", instead of "what is right or wrong". Usually when it is related to some concept I need to understand, I would alternate between 40% reading and 60% doing. If it is related to some theory that requires in depth knowledge to get it right(and hard to understand), I would alternate between 80% reading, 10% doing, 10% reflection. If it is a tiny task, I'll alternative between 5% reading, 95% doing. If copy pasting or it is something I know, 100% doing. If I'm reading something that fundamentally affects how I program or think, I'll alternative between 80% reading, 20% reflection. I use to be a perfectionist(listen more to myself), now I'm more pragmatic(listen more to the world). Ultimately, it comes down to what you are comfortable of not knowing, but still can get shit done and progress.

I start off skimming through a book/tutorials/how-to vetting if my idea of using XYZ holds and I ( hopefully) will not run into corner stones.

Then I would do "doing and learning" along the way.

This approach works for me, but sometimes I get a feeling of missing out something.

So I got me this new Haskell book: http://haskellbook.com for further enlightenment and want to first work through the book before applying it to project XYZ.

SoI guess to get work done I rely on the "doing and learning" approach. To challenge my brain/learn a new way of thinking, I rely on reading the books without skimming.

Depends on what your goal is. Do you want to Do or do you want to Learn? Pick one and optimize for it.

I've created the most beautiful half-finished architectures. I learned tons, but they were never close to being complete and I moved on to other things because over time, I realized that I wanted to learn more than I wanted to do. The good news is that after many years of learning, I am able to do things without having to really learn, but the bad news is I am no longer in my 20's, free of responsibility, or able to pull all-nighters.

To me, the two ways of learning are not mutually exclusive. Sometimes the reading motivates me to act in the real world, and often researching the problem at hand leads to further research into the larger context in order to make more informed decisions.