11:10
I can commonize the most of the invocation logic so where the only thing that has to be done is actually.
11:17
The moving logic can be done the same too. Then the save stack slot could be done the same too.
11:19
The register moving logic can be done the same also, becuase pretty much every architecture has registers.
11:20
So slotCount
is pointless and not needed at all.
12:18
I believe for simplicity that the JIT will be treated as a kind of 32-bit or 64-bit machine. Targets for 16-bit machines will be extrapolated from this.
12:29
This adjustment of the code requires a slight refactor with some methods. I have the allocate on entry which is good. I just need a commonized means of allocating registers otherwise in slots.
12:35
There is probably not a more elegant way to do this.
12:57
Ok, so instead of a single store, registers will be a multi-store operation. Since there can be no real known way to know how data is truly aligned or how many registers are needed to store values. So store register is given every single register. This means a list is used.