so far the MIPS experiment is not very satisfying. there is actually a lot more complexity to the code in that branch.

for building strings it used to look like

"Hello, " "player-name" xcall concat "!" concat

but now because of register shuffling there is more space taken up in the bytecode for pointing registers around as well as having to think about which registers the strings are in when concatting.

looking at the moment of just storing them in 20 and 21 and chaining it that way (and using push/pops to spill those) but this is still more compiler and runtime code than the forth machine takes up