Product measurability, parameter integrals, and a Fubini counterexample (Q1594954)

Summary: \textit{Negative results}: 1) A convolution \(\int g(\cdot-y)h(y) dy\) need not be measurable with respect to the \(\sigma\)-algebra generated by the translates of \(g\). 2) There exist a Borel set \(A\subset\mathbb{R}\) and two \(\sigma\)-finite measures \(\mu\), \(\nu\) such that \[ \iint 1_A(x+ y) d\mu(x) d\nu(y)\neq \iint 1_A(x+ y) d\nu(y) d\mu(x). \] \textit{Positive result}: A function of two variables, measurable with respect to a product \(\sigma\)-algebra \({\mathcal A}\otimes{\mathcal B}\) and partially measurable with respect to \({\mathcal A}_0\subset{\mathcal A}\) and \({\mathcal B}_0\subset{\mathcal B}\), is \(\mu\otimes\nu\)-almost measurable with respect to \({\mathcal A}_0\otimes{\mathcal B}_0\), for \(\mu\), \(\nu\) \(\sigma\)-finite measures on \({\mathcal A}\), \({\mathcal B}\).