So... I'm not sure if anyone has had this idea before but I finally decided to make an rpg using it.

Basically it's a grid system that allows me to make my maps infinetly huge with as many locations as I want using death counters.

First, make an area of 24x36 tiles, and make a location covering each tile.
Make X and Y coordinate death counters.
Then I make triggers that move my hero character to the opposite end of the field when he gets to the edge.
When it does this it modifies the x and y coordinates accordingly.
Then, I remove all the extra units on the field and create new ones based on where my character is supposed to be.

It plays a bit like The Legend of Zelda.

Positive: Infinite area and locations.
Negative: Can only really be done in single player, boring repetitive terrain.

What do you think?

I've experimented with this sort of thing a bit (and have seen experimental maps that others have made based on similar ideas)...there are ways to reduce the drawbacks, though they won't be fully eliminated:

-To make the terrain more interesting, you can have more than one area that you use, so instead of reusing one piece of terrain over and over, you have several different pieces of terrain to reuse. Alternatively, you can hybrid-ize this system with regular terrain, by having areas that demand higher complexity be pre-made, and more "generic" areas handled as you have described.

-For multiple players, duplicate the terrain or terrains as many times as there are players. When a player moves to an unoccupied area (keep track however you like, by constantly comparing coordinate counters, or by having a physical "map" somewhere with a square corresponding to each area, or by another system of your choice), move them to an unused copy of that area's terrain; if they move to an occupied area, move it to whichever copy of that terrain the player occupying that area is using.

Non-geometric or non-linear terrain (the terms I've always used, interchangeably, for any type of terrain where the spatial relationship between two points in the game's universe is not based on their relative positions on the map, but is, rather, under the influence of some triggered system for keeping track of the geography) allows your map to be very big, but keep in mind that a big universe, especially ones with parts being kept track of and generated/regenerated by triggers, is more work than a smaller one. The system only becomes worthwhile when used in a very large project, or in a project with lots and lots of grid (or other complex system) usage such that it can only really handle a small area at a time within SC's limits. Any map like that is most likely to be an undertaking of significant magnitude, which is probably why we don't see these concepts being used more often.

Problem is the repetition and the boredom because everything might look the same because you have to use units as terrain objects like a wall...

This reminds me of the Puzzling where, because of the Center View trigger spam, the map worked the same way, albeit still within the normal SC map size limits.
What I really was impressed with was when you would get like just a corner of a screen providing a path going between 2 adjacent screens, but completely cut off from the rest of this subject screen. It got to be quite confusing as a maze. There was also some EUD Keyboard Control RPG that used the same concept and the forest maze part was absolutely epic, you would genuinely get lost and walk in circles trying to advance through that forest.

I think Puzzling's system where all players must ALL go to the same "door" to change screens/areas is best, that way you are interacting with the other players, co-operating with them, and it doesnt get boring as easily.

Yeah The Multiple Spaces thing doesn't work all that well since the area I'm using takes up half of my locations already.
I'm using Data Discs to represent walls and mineral formations as water
The bridge I made looks really funky... and there are areas that need to be redone because the proportions of some areas are wayyyy off.
The last time I tried to make an rpg I hit the location limit so fast it was rediculous.
On the positive side the rest of the game system will work so much better this way that it's almost worth all the hard work

I'm surprised you didn't list a negative as "extremely complex and tedious trigger work required".

You can reduce the number of locations used by utilizing location grids. If you build it with a hybrid grid, with each area lined up along the unit axis, you can have additional areas with only a unit cost, and no location cost.

with ~40 different isometric boxes, you can do a lot more and make it a lot better.

This
is why I didn't do this...
or this....
Imagine using location grids to place 40 different units.
Now imagine having to do that for hundreds of squares
NOT FUN.

Unless I'm not getting the purpose of this concept, wouldn't it be easier to place a burrowed unit in every 1x1 square and just have a handful of locations focus on them? I started a single player RPG map a couple years back that was supposed to play like a Zelda or Final Fantasy Gameboy game with fixed cameras. For every frame (I divided the 256x256 map into a ton of 20x12 sections) I had a certain burrowed unit set for each direction (North, South, East, West) and a set of burrowed units scattered around the "area" for randomized encounters. Each frame recycled the same 8 locations or something, unless I needed to use more for specific purposes. The only issue with this was having to copy and paste a ton of the randomization triggers, but that wasn't too time consuming in the long run.