The Sauropod's Trunk - Introduction to the issue
The suggestion that sauropod dinosaurs might have elephant-like trunks has been around for a long time. It's mentioned in at least five books I know of, and illustrated in three of them. So, from the start, I take no credit for the origination of the hypothesis. But as soon as I read about it, I liked it. As once I liked it, I read a lot more, researched it, and have become a very enthusiastic advocate for the concept. It should be given much more attention by the paleontology community, and in the hopes of advancing it's standing as a valid hypothesis, I offer up my thoughts on the matter.
To begin, the hypothesis is born out of the unusual position of the nares (nostril openings in the skulls) of the sauropods. In the Diplodocids, the opening is at the top of the skull, between the eyes, and the opening is flush with the skull's surface in a non-descript fashion. In the Brachiosaurids, the opening is enormous, has a high arching delicate bone in the center along the medial line, and there is a massive concavity or recessed hollow in front of the openings and directly above the teeth. As different as they both are, they have one fundamental thing in common. In both cases, the nares are at the top of the skull instead of the normal pattern of being located at the front of the skull.
In the matter of comparative anatomy with existent species, no existent skulls are identical, but two types do have nares at the top of the skull. One group are the cetaceans (porpoise and whales). The other are the proboscidians (elephants).
           Comparative Anatomy
Starting with the cetaceans, comparisons were made when the sauropods were thought to be aquatic and far too heavy to be able to function on dry land. So let's look at the cetaceans and other aquatics.
The cetaceans are fully aquatic mammals which have ceased all ability to function on dry land. They are totally adapted to a life in water. But they retain the mammalian feature of breathing air. The nostrils serve this function. But in the water, swimming along, the air is only above them. It is not in front, beside, under or in back of their heads. So it seems reasonable to conclude that evolution caused the nostrils to shift to the top of the skull because this is where the air usually is. So we could say that for cetaceans, the location of air to breathe became very directionally specific, and the nostrils re-oriented themselves in that direction.
Can we say the same thing about sauropods? If you cling to the aquatic perception of them, the comparison might be valid, but as soon as you put them on dry land, as the mainstream paleontology views do today, the cetacean comparison ceases to be valid. Terrestrial creatures have air all around the head, front, back, sides, above, below. So nostrils don't orient themselves in the direction of the air. Any orientation for other purposes satisfies the need to reach the air.
If you try to make a comparison of sauropods and cetaceans on the basis of convergent evolution (independent evolution of similar features due to common or similar ecological influences) you'd be skating on pretty thin ice. A family of fully aquatic carnivorous mammals don't have much in common with quadrupedal terrestrial herbivores. So we will set aside the cetacean comparison as dissimilar to the point of being irrelevant.
That takes us to the elephants. They are quadrupedal terrestrial herbivores, the land giants of their time (now). Their forward limbs are incapable of assisting in food gathering. Their nostrils enter the skull in a flush opening at the top of the skull between the eyes. But the nostrils break the skin at the tip of the prehensile trunk at the front of the face.
The sauropods are also quadrupedal terrestrial herbivores. They were the land giants of their time. Their forward limbs were incapable of assisting in food gathering. Their nostrils enter the skull at the top of the head. Where the nostrils break the skin is uncertain. But given all the similarities above, arguing a trunk on the sauropods through the mechanism of convergent evolution is plausible.
          What Do Noses Actually Do?
But before we can make any conclusion, we need to examine the nose itself, what it does, and why it is configured on the face as it is. To begin, what does a nose do?
1. It is a breathing passage to allow air into and out of the lungs.

2. It warms and humidifies the incoming air to be an appropriate temperature and humidity for keeping the delicate lung tissue moist and warm.

3. It usually contains the olfactory sense receptors and thus is the primary device for evaluating airborne odors of all types.

4. The olfactory sense aids in food recognition and taste perception. As such, it should be situated close to the mouth. It should be situated at the front of the face so the head can point the nose at a source of odor, and ideally, the eyes should be able to see what the nose is smelling so that the brain can get a visual/olfactory correlation. With food sources, it can thus learn to see foods from a distance where odors cannot travel, and this assists it in locating food.
5. The olfactory sense aids in social identification of friendly individuals of the species. Again, nostrils positioned to permit seeing what is being smelled allow for the visual/olfactory correlation and the friendly individual (infant, parent, sibling, etc) can then be recognized by sight when odors do not carry over distance or other smells clutter the olfactory perception.

6. The nose can be used to generate sounds for communication, whether friendly or hostile.
These design criteria are reasonably consistent for terrestrial creatures. Of the six, items four and five argue very strongly for a specified position where the nostrils break the skin. And that position is at the front of the face, in proximity to the mouth, and where the eyes can see what the nose is smelling. Because of this, perhaps it is not unusual to see that almost every terrestrial vertebrate has nostrils at the front of the face, close to the mouth, and where the eyes can see what is being smelled.  
            If The Sauropod Nose Is On Top Of The Head
Taking that configuration as the standard placement with proven effectiveness, if we choose to accept sauropod noses to be where the nostrils enter the skull (on top), then we must ask, why did this creature deviate from the standard, what advantage was conferred, and how was the loss of standard uses compensated for.
1. There is no advantage in placing nostrils at the top of the head with respect to effective breathing. No advantage is conferred by the re-arrangement.

2. No advantage in warming and humidifying the intaken air is gained by the re-arrangement.

3. No advantage in olfactory perception is gained by the re-arrangement.

4. Ability to smell food sources and see what is being smelled is lost in the re-arrangement.

5. Ability to smell other individuals and see who is being smelled is lost in the re-arrangement.

6. The ability to generate sounds is not improved by the re-arrangement.
Placing nostril skin openings at the top of the head thus loses two vital sensory abilities and confers no advantage to mitigate the loss.  It should also be noted that these design constraints are for where the nostrils open at the skin, not where they enter the skull. And the skull opening and skin opening need not be in the same location.
So, at this point, we have no plausible reason why the nostrils should break the skin at the top of the head, several reasons why they should not, and no requirement or compelling reason why the skin openings must be proximate to the skull opening. 
          Why Should The Nostril Openings in the skull shift up?
Now let's examine why the nostril openings in the skull might be shifted away from the front of the face, when that is where the nostrils should break the skin.
One of the most fundamental design criteria for eating is the ability to harvest and swallow the food. Teeth are for holding or tearing apart the food, but something is needed to assist in the manipulation of the food. One option is a tongue. The second is a prehensile lip, and its specialized extension, the trunk. The tongue has no potential to effect the position of the nostrils so that is not an issue now. But the lip is situated between the mouth and nostril opening. So what happens when the nostril opening moves farther away from the mouth? Essentially it makes more room for a bigger lip with more muscle attachment area.
Intriguingly, a trunk has considerable need for just such an enlarged space between mouth and nostrils. The trunk becomes a prehensile organ with ability to pick up and manipulate food into the mouth. And it's most powerful muscles are those running on its ventral side (the underside, toward the belly) because these muscles curl the trunk around something the trunk wishes to lift or manipulate. And these muscles, running from tip to skull, need an attachment point on the skull between the mouth and nostrils. Generally, to enlarge a muscle and increase it's power, nature needs to enlarge the surface attachment area. If there is a trunk, enlarging the skull area between mouth and nostril openings in the skull increases that area of muscle attachment, and thus increases the potential for greater size and power of the primary trunk muscles that close around an object and lift it.
So an enlarged space between mouth and nostril openings in the skull argues strongly for a trunk, not simply because it's similar to the elephant's anatomy, but because it is fundamentally advantageous for a trunk on any species. There is no other explanation I am aware of for why a skull with such an arrangement might be so configured.
          Fossil Mammals with Presumed Trunks
In mammalian paleontology, whenever high positioned nares are found in a skull, a trunk is presumed. And particularly, in the somewhat camel-like MACRAUCHENIA, as described in R. Savage and M. Long's book MAMMAL EVOLUTION (pages 164 and 165), it says: "The cranium is short and the muzzle long with the nasal opening high on the roof of the skill (as in sauropod dinosaurs). Macraucheniids must have been very strange beasts - camel-like with a proboscis."
Interestingly, in mammals, the presumption of a trunk is freely given to any skull with high nares. Yet the same presumption is withheld from dinosaurs. Bakker argued that he had not seen proof that the reptilian dinosaurs had a lip sophisticated enough to evolve into a trunk. This is odd when he argues so strongly on other points of anatomy why dinosaurs are not really reptilian at all. And I side with the principles of convergent evolution that frequently let mammals, birds, reptiles and fish develop specialized similarities when the behavior in a given ecological niche is common. So seeing a common mammalian adaptation on a dinosaur is something I don't have trouble visualizing (especially when the adaptation is not predicated on the fundamental mammalian traits that define their difference from birds or reptiles). And the prehensile hydrostatic organ is not exclusive to mammals, or even warm-blooded vertebrates. Octopus tentacles are equally versatile and manipulative hydrostatic organs. 
               Concluding Analysis
Other authors, when discussing the prospect of a trunk, especially in brachiosaurus (which has this massive indentation above the maxilla and leading into the nares), these authors cite other explanations for the indentation and what possible organ it might house. Often mentioned are a blood cooling device so the brain is not engulfed in blood too hot, a vocalization device, and an unusually specialized olfactory device. But none of these things argues for why the nares are at the top of the head. None requires a shift in nostril openings into the skull from the standard front facing position. And none of these ideas interferes with or is incompatible with a trunk. You can have any of these, or all three, and still have a trunk. So there is no need to "choose one" to the exclusion of the others.
Another argument for the trunk is the simple fact that the creatures are huge, the heads are small, and herbivores need a higher physical volume of food than carnivores because the vegetation has less nutrition, pound for pound, than flesh. This all argues favorably for some assistance in gathering leafy matter and shoving it into the mouth as fast as possible. Given the arms cannot do so, and the tongue can only do so marginally, a trunk would be the most helpful device to feed efficiently
So, adding it all up, my vote is most emphatically yes to the trunk hypothesis.  To see how it might look, click here
Updated Commentary     Diplodocus Discussion      Brachiosaurus Discussion