Yeah, it's kind of ironic that the modern - and entirely rigorous - direct treatment of infinitesimals comes under the name "nonstandard analysis."

As to the foundations, what Fourier's work shattered was the sufficiency of then established notion of what a function is, which had been limited to what we now call analytic functions; this eventually lead to the abstract definition of a function that we have today.