change number representation

[speedy-lex]

Represent numbers as *char but store the actual bits in the char

Currently numbers are being stored as strings (eg [‘1’, ‘2’, ‘3’]=123) which takes more time to do calculations with that to be storing it like this [0x1, 0x24] = 1*(16^2)+2*(16^1)+4*(16^0).

Why *char?

I suggest using char because it is exactly 8 bits on all machines so the code is (mostly) the same for 64 bit and 32 bit machines.

[avighnac]

Hey, this change makes sense, but I've pretty much completed this project ATP, and changing something so major would take a lot of time and effort.

Interestingly, I was also experimenting with larger bases (2^32 or 2^64, which I think would be even better, or 10^19)

Anyway, one of the reasons (not that this justifies the slower speeds, just thought I'd state this for the sake of it) I chose to go with base-10 is because it was the easiest to code and debug. Also, when read in base-10, the operations would be precise to the exact decimal place you want them to be (which, like, yay!)