[mkl]

I prefer Python to Octave and Maxima. Numpy, scipy, and matplotlib for numerical stuff, and sympy for symbolic stuff. Having them together in the same general purpose language is really convenient, and Jupyter notebooks are fantastic. Sage is also good, but I've moved on to sympy. I don't know a way to get step-by-step working from a library, but sympy gamma can do some, so it's probably possible to some extent.

My experience suggests avoiding Maple like the plague. Sympy (and Sage) can do everything I ever used it for much nicer and easier.