Not to offend or insult but...
I hear the phrase "I can't see any new classes for D&D. The current one do pretty much everything." in some form or another in D&D forums a lot.
And I wonder how creative or imaginative the fanbase is. Is it:
"No other classes fit in D&D"
"No other classes fits in my image of D&D"
"I can make almost any other class using what we have."
or
"I just can't think of any other classes."
The first two are a bit narrow minded with a bit if selfishness in the second IF the person isn't willing to support others at least via encouragement or constructive criticism.
And although some classes of the past are shallow or redundant in mechanics or flavor concept in the past, it would need little much more than reimagining of the class. That's what they did for sorcerer, ranger, paladin, and warlock. D&D has a few classes like that. They have something unique and only new the other part.
I'm picky with new classes because I think classes are big tentpole options. They're huge. They have to be something that influences your character from level 1 all the way to level 20.
As classes can be played from level 1 to 20 they also require an eff tonne of playtesting. It's very easy for classes to break things since they affect every round of every combat for an entire campaign. Because of the time requirements alone, new classes should be rare as you can only test so many classes at a time.
Classes should also be a marriage between mechanics and story. You can have a story based feat or a mechanic based feat and those work, because they're small. But a class needs to be more than a neat mechanic or a fun story hook. It needs to have both.
It's super easy to think of new mechanics. That's not an issue. But a class needs to be more than a veneer of story and lore over a cool mechanic.
Similarly, a character concept is not a class. Classes should not depend or assume a particular backstory or origin: that's character fluff not class fluff. I was super hard on the warden for this during 4e, as part of it's "fluff" was someone being raised by bears, which is a super fun character concept but a terrible class concept. A bear cub human could easily be a ranger or barbarian or druid. It's not distinct to a class. (And it doesn't work in a world without bears.)
Classes also need to have some variety. A one-trick-pony makes for a poor class. I've said this for Pathfinder, as a class needs to be big enough to encompass the one concept and archetypes. Similarly, a class for 5e needs to have lots of mechanical and narrative room for subclasses.
D&D books should also have some level of genericity. Almost everything published should fit into the Forgotten Realms and Eberron without issue, as those settings are meant to be kitchen sinks. If they don't fit easily, the content might need work. Similarly, new content should be stuff that can easily be added to a homebrew world without work. For that reason, new subclasses are always going to be more desirable as they just augment classes that have a role, a place in the world.
Brand new content that is funky and doesn't easily fit in a world is much harder. Like Magic of Incarnum or shardminds. They're neat ideas but really require some lore buy in and are hard to incorporate into a fantasy world in play, or a campaign setting that might have been the setting for games since 2e. Anything that big is something that just doesn't work as something that just hasn't been encountered before, as it's dramatic enough that it should have impacted history in some way.
Because of the work required to design a class and make it balanced, new classes shouldn't appeal to a small fraction of the community. Because then the work was wasted. The more content you release, then the more those wasted classes really stand out.
Let's say there are two-dozen classes and one class is odd and only half of all groups allow it. If all classes are equally popular, and groups have an average of 4 players, then one in six groups will have that odd class. So 15% of 50% or roughly 7.5% of all tables. And that's assuming all classes see equal play, and no one doubles up. Making a new niche class is a lot of work for something 93% of gamers might never see in action.
Even obscure as those are, some got treatments. The swashbuckler and Beguiler got PPs and dragon articles, PPs made a small attempts at incarnate, healer, themes or backgrounds could suggest swashbuckler, healer, archivist, and there was a wilder theme, duskblade was just another 3.x stab at a gish, and 4e also had plenty of those, any arcane striker could probably claim 'warmage' (it was another mechanics-differentiated class)...
... I wouldn't expect any of those to have been satisfying, but given more than 2 years (ie no Essentials re-boot), I'm sure they could have been tackled better eventually. And, of course Pathfinder was there within a year to continue those classes and add ever more, as well.
So, even though there's no expectation of 5e ever being as option-rich as 3.5, it's still no comparison, really.
Some of those were from the Player's Handbook 2, so they should be no less "obscure" than the avenger, invoker, shaman, or warden.
Given late-4e was willing to do stuff like the runepriest and seeker rather than update 3e content, it seems highly likely they would never have updated those 3e classes. Just like they opted to make something completely new with the swordmage rather than go to the bladesinger, duskblade, or eldritch knight for inspiration.
After all, 2e was arguably the longest lasting of all editions, and it didn't really bother adapting the classes from 1e (or just did them as kits).
As I said earlier, every designer wants to make something new and leave their mark on D&D. Everyone wants to make the next great iconic thing like the warlock or Far Realm or chuul that moves from edition to edition.
Within 10 months, 4e had every class that had appeared in a prior PH1, plus a few from more obscure sources, plus some completely new ones.
We're over a year into 5e and it hasn't even covered everything that's been in past PH1s.
Excluding the assassin, which was in the 1e PHB but took well over a year to appear in 4e Dragon, and even then only as a playtest.
Really, 5e is pretty good, with the assassin and warlord being covered as subclasses. And it has everything in a single book rather than delaying some classic content for the better part of a year.
4e, like 3.5e, was also super heavy into the splatbooks. So it seems like an unfair comparison. That's not where the focus is.
Within 13 months of launch, 5e has three complete campaigns running from levels 1-15. No edition had anywhere close to that. It took 3e two years to go from
Sunless Citadel to
Bastion of Broken Souls, and a year and change for 4e to move adventurers from the
Keep on the Shadowfell to fighting the
Prince of Undeath.
If we got creative enough (and were charitable enough about meeting concepts with very tenuous modeling and depending on lots of generous DM rulings), we could probably build any concept with MC'ing and a Fighter, Cleric, & Magic-user class. (Even the Rogue would just be a light-fighter with a background and a lot of skills.)
It'd take that level of squinting and settling to do a lot of concepts with the current classes. 5e devotes a lot of sub-classes (~30 out of 38) to doing this or that caster-concept (some of them little more than meta-game concepts) just so. It could do with more classes and sub-classes to handle other sorts of concepts.
Subclasses are the big thing.
Builds in 4e were rather small and more of a theoretical design, since you could take powers designed for the build or choose other powers entirely. There was less design space as the build effectively came down to a single small class feature: there was less room to make something brand new.
In 5e you don't *need* a new class to same extent as you did in 4e or even 3e, since some class flexibility is baked in. So alternate concepts can more easily become a subclass, or even a variant class (like the spell-less ranger). So a spirit shaman can become a druid build, a shapechanger can be a barbarian or ranger, a soul knife can be a monk, and so on.
