Handling of invalid opcodes & C translation

[RockBrentwood]

The "invalid opcode" message may cut off the assembly code generation in mid-stream. A better way to handle it is how DEBUG.EXE originally handled it: which is to treat it as a byte constant, e.g. "db 0fh" for opcode 0f. So, "db" should be added as a pseudo-mnemonic and the table entries set to that, instead of to "nop".

There's no reason to treat the floating point escapes specially - their format is standard, essentially the same as opcodes 84h or 85h (to cite the two most typical cases). In octal, the opcode sequences is \33D \xrm [ExtraBytes], which is "escape D" with (nominal) "register" r=0,1,2,3,4,5,6,7 and (nominal) "address mode" (x,m) and "ExtraBytes" dictated by the values of the octal digits x=0,1,2,3 and m=0,1,2,3,4,5,6,7.

Similarly, "aam" and "aad" do not really need to be treated specially, but could be treated the same way as "int P" (opcode \315, i.e. 0cdh).

I translated what you have - without any of the above-mentioned changes - straight into C. I'll fork off of your version, add the C translation, and then do continual upgrade on it from there. You're welcome to copy back the C translation (or any of its upward revisions), when I put it up.

Edit: 2023-07-18: I forked and put it up on GitHub.