@hamishknight One more thought while we're on the topic of whitespace: since the proposal already proposes rejecting bare /.../ regex literals with leading spaces, is there any reason to think it would be unduly restrictive also to reject bare /.../ regex literals with trailing spaces?
Besides restoring a symmetry which is subjectively an aesthetic improvement, since most idiomatic styles use spaces surrounding binary operators and after commas, this simple modification of an already proposed restriction would eliminate here even the need to disambiguate foo(/x, y / z) at all, as well as the need to disambiguate foo(/x, /y), or foo(/, /), etc.