Darkvision
my (fairly large) two cents:
I'm inclined to agree that as written, darkvision is "I can see...in the dark!" There are no funny loopholes, no complications, and no physics involved. You can see when the lights are out. You see black and white. End of story.
However, if I were to modify things to suit some kind of sensible physics (as many are wont to do), I would have to ignore the RAW and change things a bit. Normally, a human can see colour and intensity. When it's very dim, the colour vision (cones) shuts down for lack of light and the intensity vision (rods) takes over, which is why in very dim light you can't see colour (everything generally looks bluish). The only reason why you can see degrees of intensity in dim light is because there is some light reflecting off the object.
Now, if a dwarf or other darkvision-possessing creature sees in the dark something that is simply a black-and-white version of what they'd see in the light, certainly they'd be able to read ink on paper. But when such a creature looks at something, that creature is not going to see a greyscale like the one that a normal-vision creature would see in dim light. That greyscale is a product of the way light reflects off the object. If the object is not reflecting light, which is the assumption in a darkvision situation, there's no reason why there should be a greyscale of the sort a normal-sighted creature would see.
Now, the picture in the old DMG, as you'll notice on a close examination, is not simply a greyscale version of the normal picture. It has been greyscaled, inverted, and an effect was run that emphasizes the outlines of things so that texture and depth are exaggerated. The overall effect is that smooth surfaces and empty spaces are dark, but the edges of things and changes in depth are lighter. This suggests that what darkvision picks up is changes in depth.
Still, if you can see anything at all and the ability isn't just magic, there has to be some kind of vector carrying the information from the object to your eyes. Since there is no light, you must be producing that vector yourself. Even owls, who have the best night-vision in nature, cannot see in total darkness, but we are told that a dwarf can. Therefore, the dwarf must produce either a sonic or radiant emission that reflects off the objects he sees. We know it's not sonic, so it must be radiant. We can assume it comes from somewhere on the dwarf's face, probably on or near the eyes themselves (though not from inside the eyes, because that would blind the darkvision receptors), and that it projects this radiation in a cone or hemisphere the way a flashlight would. A hemisphere is more likely, to allow for peripheral vision, but I'll continue talking about cones here because it's easier to type.
Since this radiant emission is what makes darkvision possible, it must be detectable by other creatures with darkvision. However, like a light-source, you would only be able to detect it on the surface of objects, perhaps on dust in the air, and when looking directly at the emitter. But let's say dwarf #1 comes up behind dwarf #2 to surprise him. Does dwarf #2 see dwarf #1's darkvision cone on the wall in front of him? Maybe, or maybe it doesn't have an additive effect, and since his own darkvision cone is already on the wall (since he's looking at it), he wouldn't see the addition of a second cone. However, should he turn around, he will see dwarf #1, who will likely have a bright patch at the location of the emission of his darkvision cone.
Or maybe not. If the darkvision emission is not additive, then dwarf #2's darkvision would dampen dwarf #1's darkvision emitter, and dwarf #2 wouldn't be able to see it unless he covered his own emitter, which would effectively blind him to anything that wasn't in dwarf #1's darkvision cone.
The question of whether a dwarf could see through glass or water is a simple one that amounts to the question of whether glass and water can transmit the darkvision radiation. If they can, then the dwarf can see through water. If they reflect it, then the dwarf cannot. While a simple question, this isn't an easy problem to solve. Normal glass transmits visible light, but not infra-red light. X-rays, which are the same kind of energy as light (electromagnetic radiation), can travel through objects of low density, but not through objects of high density. A hypothetical darkvision radiation, which may or may not be electromagnetic in nature, might pass through glass but not water, or water but not glass, or neither, or both. Note that both water and glass are opaque to visible light across large distances, and that the angle of incidence of the light ray will determine the percentage of photons that are reflected by or pass through the water or glass.
Back to text: text should only be readable if the darkvision emission reflects differently off objects that would reflect light differently. Black ink tends to absorb photons, and white paper tends to reflect (and scatter) them. There is no indication (because we're making this up) that the darkvision emission would respond to ink the same way light does, for the same reason as was given for water and glass, above. Which means, again, that it's DM fiat whether this radiation is absorbed by "dark" objects and reflected by "light" ones. However, going by the picture of the mind flayer, we can guess that the darkvision radiation responds differently than light does when encountering objects. Otherwise the darkvision picture would look just like a greyscale version of the light-vision picture. Since it does not, we can assume that the darkvision radiation ignores "dark" and "light" objects...in other words, it behaves differently than photons do when encountering objects.
In fact, it seems as though darkvision is absorbed by smooth surfaces (giving black areas) and reflected by edges. However, this doesn't make much sense physically. Assuming that the object is opaque to the emission, an edge should deflect it and a smooth surface should send it directly back, exactly as a billiard ball changes direction when it hits a corner and bounces directly off a flat surface. Also, normal light scatters on a surface, but darkvision radiation does not seem to. An explanation for this problem is that the darkvision emission does in fact reflect off smooth surfaces and scatter on edges, but the dwarf's eyes are set up to interpret high radiation intensity as "dark" and low darkvision intensity as "bright". Still, an empty area must seem dark to the dwarf. So we'll amend by saying that the edge of a high intensity area is "bright" but both high and low intensity reflection areas are "dark". So a dwarf only actually sees the edges of the areas that reflect darkvision radiation. This explains the smooth/rough problem.
So a dwarf can see through water or glass if the DM decides that the darkvision emission can pass through water or glass, because there is no other logical or textual basis for the decision. But it is entirely reasonable to say that dwarves cannot see written text in the dark because the ability to differentiate intensity is a feature of light-based vision, unless the ink absorbs darkvision radiation. It is not unreasonable to guess that dwarves would have discovered some substance to use as ink that either left a raised enamel on the paper that would be visible to darkvision, or that absorbed darkvision radiation, thus creating an artificial "edge" that would be visible to a dwarf. This would create the visual effect of the words being holes in the paper with nothing behind them, which is actually kind of neat.