Show the actual value of the flags registers if there is a mismatch in the lower nybble

[yoursred]

No description provided.

[robert]

Ahh do you mean show the full 2-digit hex, rather than just the 4 flags? In case there's a mismatch in the bits in the F register outside those used for the flags?

[yoursred]

Yup. Only show it if there is a mismatch in the lower four bits.

[robert]

👍 makes sense, I'll do that at some point!

[CosmicBagel]

This sent me for a loop, I had to manually look at the logs to figure out what was going on. F register was getting bit 0 set, gb dr said the line was wrong, but the diff it displayed me was identical. Specifically happened for the 01-special.gb test.

[robert]

Thanks for the vote, I'll keep this in mind. If you're interested in submitting a patch then that would be awesome, although the code is a little gnarly so might be tricky XD.

[dan-mcdonald]

I think the lower nybble of F is always 0.

ld b, 0xff
ld c, 0xff
push bc
pop af
push af
pop bc

If you check c after all this in an emulator like BGB, the lower nybble will be zeroed out because when the values were pushed back onto the stack the lower f nybble is zeroed.
The cpu_instrs does something goofy like this to feed state to the CRC checker and I found the lower nybble was zeroed out.
Also this reference shows the lower nybble as zeroed out.

I'm not an expert though. If I'm wrong, please don't neglect your duty.

[robert]

Heh you're probably right. Do you think we should change anything about gameboy doctor or is this just interesting trivia?

[dan-mcdonald]

I just commented to be helpful since it sounded like a fellow GB emulator developer was on the wrong path, and people from the future finding for this issue might benefit.

I think the ideal output from gameboy-doctor would be to check the lower F nybble and when it is non-zero explain that it should never be zero.