In an old paper from 2003, D. Aharanov simplified Shi's proof that CCNOT (or Toffoli) gates + Hadamard gates together form a computationally universal gate set for quantum computing. The CCNOT gate is well-known to be computationally universal for reversable classical computing; another similarly well-known such gate is the CSWAP (or Fredkin) gate.
Can we plug in the Fredkin gate into Aharanov's paper to prove the universality of CSWAP + Hadamard for quantum computing?
In her proof of the universality of Toffoli + Hadamard, Aharanov considers $\Lambda(U)$ as a controlled version of $U$ - she writes the Toffoli gate as $\Lambda^2[X]$. Does something similar work for CSWAP?
Toffoli + Hadamard is an interesting computationally universal quantum gate set, at least because there's never any imaginary amplitudes (unless provided by ancilla qubits). The Fredkin gate is an interesting computationally universal classical gate, because the one's count (Hamming weight) is conserved between input and output.