> Armchair physicist here

I've started and aborted a couple of ELI-n where n < ~ 24, and I'm not sure this one is especially satisfying either, but I think my previous attempts were just confusing and went beyond your implied question.

Causality in a sandwich wave spacetime is weird. All the null geodesics from a point can eventually refocus on some future point in the afterzone. We can take that simple fact (from Penrose initially) and contrive a sandwich wave spacetime that shows off the Bondi&Pirani words that confused you.

If you do an isotropic pulse of light in the beforezone, as expected (beforezone being just the flat spacetime of special relativity) it flies outwards in an expanding spherical shell. In a flat spacetime without a sandwich wave, you'd never see any of it again in the absence of a reflection. However, in the sandwich spacetime the gravitational wave can roll over you (and this expanding shell of light) such that in the afterzone all the light that you flashed out isotropically, eventually lands all at once on your eyeball, no mirrors required.

Moreover, the null geodesics out of multiple different points can all be refocused on just one future point. So you can fire off dozens, millions of isotropic flashes in the beforezone, and fry your eyeball (in the future, in the afterzone) to a crissssp.

Going further, one can set up a sandwich spacetime so that the light from every event in the beforezone lands on one point in the afterzone, or on a succession of time-ordered points. In this way, an observer in the afterzone can in finite time receive light from every light-emission in the beforezone, even if the beforezone is infinite.

Finally, I am using "light" loosely; to keep the sandwich spacetime flat except for the wavezone, instead of light we need massless test particles that feel gravitation only passively. They don't exert an active influence on the gravitational wave or on the focal point in the afterzone, even if there is an awful lot of them. More technically, we need the stress-energy tensor to be 0 everywhere in the whole (vacuum) sandwich spacetime, otherwise we will have a lot of work[1] exploring the contrived spacetime's insensitivity to perturbations from matter (which generates curvature after all, and a heavy enough amount can distort the wavezone). Matter here includes light.

- --

[1] there is a literature in that area, with Mustafa Halilsoy being fairly prominent in studies of (matter) waves including standard electromagnetism interacting with the sandwich (gravitational) wave, which in turn invited thoughts about (non-sandwich) gravitational waves interacting with the sandwich wave, which is also in the literature. I'm not likely to ELI any of that here though.