Maybe I could have it so the characters completely regenerate health with each new battle, like I was going to do, but have it where if the character dies, you lose them for good and have to choose another character from your roster for the empty slot.

This implementation sounds more like resource management. Further explanation, if you replaced a dead character with another which could perform the same or similar duties then it will not have the full impact of a permadeath.

A permadeath would imply that when a character dies, not only are they never coming back, but the player has lost something they can feel emotional about. Possibly ending the game right there. It is popular among hardcore players who like punishment or extremely unfair challenges.

Well, as I understood it the characters will level up and if they die you will lose a high level hero.

A friend of mine had this idea for a gladitorial combat sort of RPG once. The idea was something like you had a number of fighters under your "employ" that would engage in various battles which may or may not be to the death. The battles could be 1 on 1, free for alls, or vs some critter of some kind. The whole point was to generate ticket sales and prestige. Batlles involving fighters of higher levels and popularity of course would be a bigger draw. Ultimately you could expect veteran fighters to loose due to injuries or an unfair battle but you could also discharge the veteran if you thought it'd be to your advantage (maybe you get a temporary bonus to recruitment).

Since the game was about pretige and ticket sales, loosing high level fighters wasn't as serious a setback. People might pay big money to see the young upstart that got away with a lucky strike at the reigning champ.

The only thing is, once you start looking at money as the player's measure of success, you're going to start thinking about things you can do with that money. And that could very quickly explode the scope of your project into something you don't intend.

I think that when it's correctly done, permatdeath is a very good thing. It can create some emotional link between the player and his soldiers/creatures/characters, and encourage the player to take care of them. But the death must absolutely be justified. I wouldn't like my character dying for nothing or randomly.

Good exemple include Fire Emblem of course, or also Final Fantasy Tactics : Lion War. It can also happen in some STR. I was for exemple very attached to Bob a Giant Fly in Dungeon Keeper. (True story, it was his randomly created name!) I trained him, I possessed him to go through my dungeon, etc.. But he fell in battle. Okay, the Giant Fly is the weakest creature of the game, but it was Bob!

What I want to say, is if a player spend time with a unique, irreplaceable character, and if this character appeal to him, and not only because the character is strong (good look, good feeling, funny details), there is a chance that he will be affected by its death, and more attached to it. And the fight will be more intense.

But It doesn't works in competitive games :)

And yes, I gladly sacrifice efficient character for "character with whom I have a fell"

A game that is nothing but one turn-based battle after another is often included in a social gaming site. Flight Rising, for example, has The Coliseum which is this kind of thing. They aren't my personal favorite type of combat to put into a virtual pet site or similar social gaming site, but the owners of the sites are buying them/hiring people to create them, so there's a market.

So what if I had a game where the game is absolutely nothing but turn-based RPG battles, one after another, and long ones too?

That's a great idea! The game Gratuitous Space Battles was built in the same manner, just focusing on the combat.

So what if I had a game where the game is absolutely nothing but turn-based RPG battles, one after another, and long ones too?

The key to making this work is variety - not necessarily content wise, but in moment to moment conflicts. Do not just have long battles, have lots of short ones too. Vary the number of monsters you fight against - have some battles with a couple strong enemies and other battles with several weaker enemies. Make each battle feel different than the last. Perhaps the player must handle each battle differently depending on the enemies present, the environment you're in, or the characters you currently have at your disposal. That's another huge opportunity for variety - having the player control a rotating set of characters from battle to battle. Each character should have a unique fighting style (like in FFVI) which forces the player to handle each one differently.

Rather then focusing on how to reward the player per game, focus on the turn-based battles themselves. In reality, there is no real way to reward the player if you're going to have consecutive turn-based battles. Sure, adding a hero that levels up might be a nice idea but as it is the only constant, the attachment to the hero would become stronger then the wish to play the game itself; having perma-death on him would actually discourage players to play the game.

To get back to my original point, focus on the turn-based gameplay itself. Turn it into something next gen, where anything seen on the battlefield can be used to your advantage; or disadvantage. Say your orc has spiked armor, any melee fighters attacking him would have damage inflicted upon them as well etc. You threw a poison gas bomb at the enemy, but the wind direction drifted it towards your line instead and severely injured some of your minions.

Don't stop there, you have an elf and an orc fighting together but you didn't know that in their history there was a huge battle that put their two races as enemy's and so they turn on each other in the middle of an attack, wreaking havoc and chaos that allowed the enemy to overcome your superior forces. You started to travel across a bridge, but an assassin had planted a bomb that would blow up when stepped on. Your forces get thirsty and take a drink from a river, but it turns out that their were enemy mermaids waiting in ambush; taking out half your forces before retreating into the distance with their superior speed in the water.

Okay, that's enough examples. But just think, a game where your strategy's can become a reality. You don't have to think about balancing it with story's or anything, just create the most versatile turn-based strategy game. Well, that's my opinion.

I was told that making my Final Fantasy VI style RPG was too difficult for one person

False

now that I no longer have so much passion that I want to debate.

That's the only reason you can't do it then.

So before I scrap the idea, I have another idea. Something I really wanted to do was implement a turn-based battle system. So what if I had a game where the game is absolutely nothing but turn-based RPG battles, one after another, and long ones too? Meaning no real storyline, no towns, no navigation through areas, etc. It would cut the amount of work in half.

Bahamut Lagoon does essentially that. The game is one tactics fight after the other with a small mini-level aboard the ship for the most part (and few cutscenes).

Most of the effort is put towards the combats, but its still a big game.

What you are doing now is called "scoping". You're cutting the design to a level you feel comfortable you're able to complete, and this is a good practice if you intend to complete it before motivation / capital runs out.

I would recommend that you make it in such a way that its easy to MOD.

For example, if battle data is sent through a known format (JSON or .xml for example) it will be easy for other people to contribute tools if the core game is fun, and it will enhance the experience for everyone with new scenarios.

I recommend having a look at battle of wesnoth.

Good luck.