moa/todo.txt
2021-10-21 21:55:27 -07:00

39 lines
1.7 KiB
Plaintext

* generics for the frontend that's passed to the builder functions is much better than a trait object
* you could possibly use a backend representation that is given to the frontened, rather than vice versa
* you could use callbacks in some way (ie. callbacks from the gui loop to the system)
* should the frontend and simulator parts be separated so that the simulator part can be a library package?
* it should be call something other than canvas because it'll have input as well
* how will you get the canvas/app shared object between the sim thread and the io thread? It kind of has to be passed through the system-creation
functions (an only if they need them), which raises questions about how to configure things. It'd be nice if you could still run the same hardware
simulated without using a frontend
* can you eventually make the system connections all configurable via a config file?
* test using mpsc to pass messages with the tty IO thread, and test if it's slower than what you have now
* make it possible to break out of the current execution, into the debugger, by using a certain keystroke
* make tests for each instruction
* unimplemented: ABCD, ADDX, BFFFO, BFINS, BKPT, CHK, EXG, ILLEGAL, MOVEfromCCR, MOVEP, RTR, RTD, SBCD, SUBX
* undecoded: ADDX, SUBX
* modify execution for >=MC68020: MOVEM
* >=MC68020 undecoded & unimplemented: CALLM, CAS, CAS2, CHK2, CMP2, RTM, PACK, TRAPcc, UNPK
* add support for MMU
* add support for FPU
* Coprocessor instructions: cpBcc, cpDBcc, cpGEN, cpScc, cpTRAPcc
* should you simulate clock ticks, and only step devices when they next request it
* how can you have multple CPUs
* should you simulate bus arbitration?
* check all instructions in the docs