I agree, but others do not.
Consider the caboose example in the OP and the responses to it. If the train is traveling at the speed of light, the light on the caboose will not shine and light up the engine train, for the speed of light would have to travel faster than the caboose, and light cannot do that because that would require the speed the light is traveling to exceed c.
The train cannot travel
at the speed of light. As an object with non-zero mass approaches light speed, adding more energy to make it go faster contributes to its mass, more than it contributes to its speed. Kinetic energy is proportional both to mass and to the square of velocity, so to add a bit of velocity to an already fast moving object you need to add a lot of energy, which means you also add a lot more mass, and as the velocity approaches light speed, the mass (and hence the extra energy required for further acceleration) approaches infinity. Unless you have infinite energy available, you can't accelerate an object with a non-zero rest-mass to light speed; and an object with zero rest mass (eg a photon) cannot travel at any other speed in a vacuum, because to do so would imply zero energy - and an object with zero energy and zero mass is equal to no object at all. Interestingly, this also means that from a photon's perspective, time doesn't exist - a photon that arises in a quasar at the edge of the observable universe, and that takes millions of years (as measured by us) to reach our telescope, will, from the perspective of the photon, have been emitted by the quasar and absorbed by the telescope's optics simultaneously, having traveled zero distance. To a photon, everywhere is the same place.
If the train is traveling at 99.999% of the speed of light (add any finite number of 9s after the decimal point), then the light on the caboose will travel at the speed of light - as measured by the guy on the caboose - and return to him, lighting up the locomotive; It will be just as if the train wasn't moving at all from his POV.
An observer next to the tracks will see the same thing - but the train will appear to be very much shorter than the guy in the caboose thinks it is, so the light will still travel from one end to the other in the same amount of time, despite only moving fractionally faster than the train itself.
If the train is accelerating, trying (futilely) to 'outrun' the light, then the observer next to the tracks will also see time slowing down to conserve the speed of light from all perspectives (and from the POV of the guy on the train, the guy next to the tracks get old a lot faster as time appears to speed up).
No matter what you do to try to make light speed appear different to different observers, each observer's measurement of length and/or time will adjust exactly enough to keep the measured light speed constant.