I'm going to go back to my standard political pitch: there is no such thing as a "long distance train".
The *so called* long-distance trains are simply corridor trains stuck end to end for lower operations costs, with sleeper service on the overnight sections for added profits.
(After looking at a lot of numbers, I am entirely sure that sleepers are more profitable than coaches on the Lake Shore Limited, and probably on most of the other Amtrak trains which have sleepers too. One should be restored to the Night Owl.)
The only long-distance train which really doesn't fit this model is the Sunset Limited, due to the vacant deserts in West Texas and in Arizona. Arguably the Capitol Limited, with unusually high end-to-end traffic, also doesn't fit this model. All the other so-called long-distance trains very clearly do fit the model of "strings of corridors", and I believe should be thought of in this manner. This does drive my support for certain proposed reroutes -- any reroute which hits a bunch of siginficant downtowns instead of hitting tiny rural towns is likely to improve the route's "corridor" character.
As for dining cars... well, you want to feed people if they're on the train too long. How much people care about this (versus brown-bagging it) seems to be regional, though. According to the last set of studies I read, namely the PIPs, the only train which gets high coach patronage in the dining cars (I think it's around half coach passengers) is the Lake Shore Limited. It's evidently the most profitable train to put a dining car on. (So Amtrak, put the dining car back already, ya dopes!) I figure this is because it goes to three areas with high restaurant prices (NY, Chicago, and Boston) -- also three cities where there are a lot of well-to-do people who don't drive -- and so I think it attracts non-price-sensitive riders more than some of the other trains.