In the usual answer to this kind of puzzle (which more typically involves running with a ladder through a barn, or something similar), the solution is that whether the ladder is inside the barn is not a Lorentz-invariant statement, because it requires information about what is happening at two points with spacelike separations (the ends of the ladder, at equal times; whichever observer's time is used, the ladder ends are spacelike separated). So it is no surprise that different observers see the situation differently.
In this case, there is an additional mistaken assumption, which is that as long as part of the bus is on solid ground, it will not fall. The goal of this modification was presumably to try to make the outcome (whether or not the bus eventually falls) clearly invariant. And it is a Lorentz-invariant observation whether the bus falls or not. In fact, the bus does fall. This is obvious in the frame where the bus appears at some point to be completely unsupported. In the frame where the bus is slightly longer than the gap? Well, you trying driving a 65-foot bus across a 60-foot gap in the road, and let me know if you make it.