In numpy, how can we detect uint32/int32 overflow or underflow?

Question

I was surprised that even setting np.seterr(all=\'raise\') doesn\'t seem to trigger overflow/underflow errors for int32, despite the documentation seeming to say that these