Up through 3e, traps remain consistent in their general form*: something for the rogue to do. With a few exceptions, the only thing other classes can do is suck up the damage, since they have no particular chance to even spot the trap. (A special exception goes to those earliest editions of D&D that had no rogue or thief class.) Anyway, traps consistently represent a lot of dice-rolling by the rogue while the rest of the team sits on their hands. On the other hand, in some editions the thief is such an undesirable class that this is just about all they have going for them. 3.x completely changes this - as long as you're not fighting things that are immune to Sneak Attack, the rogue is a wonderful class with a lot of interesting options.
Still, 3.x retains the ham-fisted approach to niche protection, making it impossible for any other class (no matter how hard they try) to equal the rogue's trapfinding and trap-avoiding abilities. Trap design could not do a lot more to exclude other classes. The next problem is that, even for the rogue, it isn't all that engaging - there's a skill roll or two, and failing any of these rolls results in harm or death. The apparent decision points are whether to advance slowly enough to check each square thoroughly, and whether to disarm traps or just bypass them (usually with varying degrees of difficulty).
Let me break this approach down just a tiny bit more. Given the setup, what's really going on is that rogues provide the party with an extra layer of saving throws against the trap. If the first saving throw (detection) is successful but the second one (disarmament) isn't, the rogue takes the damage instead of someone else, or everyone else, doing so. Like all saving throws, there aren't a lot of choices to make; unlike other saving throws, the player is expected to alter her approach to gameplay in order to be allowed the "detection saving throw."
There's a countervailing point to this criticism of trap mechanics: they're designed based on feel, nominally imitating mechanisms that make sense for the dungeon's residents to have installed, both tactically and technologically. Tripwires, pits, pressure plates, trapped locks (and locks themselves, for that matter) and so on, are relatively simple and easy to understand. I'll be coming back to this point.
*Especially in the 1e era, another approach to traps also appears, in the works of the incomparable and bizarre Grimtooth, by Flying Buffalo. Grimtooth's traps are famously elaborate, ridiculous, and unfair - the latter of which may be a reasonable tactical decision for the defenders, but is a terrible gameplay decision on the GM's part. Still, convoluted traps that endanger the whole party are not all bad - everyone is invested in and able to contribute to finding a solution. In Grimtooth's works, the insane deathtraps are often obscure, guess-what-I'm-thinking physics challenges. I hesitate to call them puzzles, on the principle that puzzles have clues. They fit tolerably well into funhouse dungeons, but most of them would feel out-of-place in a dungeon more focused on verisimilitude.Then there are glyphs and symbols. Rogues can theoretically do something about these, but it's a lot easier for level-appropriate spellcasters. Glyphs enter play when other kinds of traps are still relevant, while symbols make their first appearance in 5th-level spells, and keep improving up to 8th-level spells, long past the time that the hardest traps in the game are trivial for a rogue who has put any points into traps skills. My point here is that magic completely replaces mechanisms, and 3.x glyphs and symbols can combine with a normal encounter, or they can be an encounter. (Ahem, greater glyph of warding, spell: with summon monster VI, this becomes a single creature of CR 5-7. Maybe a little low for PCs going up against a caster of level 11+, but as a way to delay and raise an alarm, it's not too shabby.)
4e took a drastically different approach to traps (and hazards; for simplicity I'll use "traps" to cover hazards as well). Recognizing that the 3.x approach to traps generally entertained only one player, and didn't offer that player particularly interesting choices, 4e brought traps more in line with monster design (including giving them "creature" types) and linked them directly to encounters. That is, players really only encounter traps in connection with monsters. Also, dealing with traps very often requires the skills of other party members: Athletics, Dungeoneering, attack rolls, and Nature. Around 16th level, Arcana and Religion suddenly replace all of those as the useful skill of record. This follows "high level = high magic" well, but it's a bad mistake in terms of game support for all of those other skills. The bounded accuracy of D&D Next pays off well here in redeeming this error - those low-level traps could still be a hindrance, if not a threat, and thus the skills to thwart them retain this more interesting application.