Hi Robert, actually, by calculating the derivative of the antiderivative and subtracting it to the function itself, as you suggested, it returns zero, after proper simplification. Hence, I would say that Maxima's result is valid. I've been influenced by the results indicated on the table of integrals and obtained with Wolfram Alpha, so I was uncertain about what I got with Maxima, but it was only a different representation. Sorry for not checking the correctness of Maxima results more thoroughly...
wrong (?) indefinite integrals involving square root