Damn the diff is impossible to read with just one word per line. How do manpage maintainers do this?

Two comments of the things I can decipher:

• Please don't call the bit index bidx, that sounds odd. I think this was a suggestion that did not make it to the PR. It was in an email from Github but I don't know why. I'm okay with all other options, including keeping it at u3.
• The last line, "byte by default if one is not specified", sounds really odd, but I'm yet to come up with a good alternative.

Thanks for your feedback!

• My own preference is to keep u3 since it's an accurate spec of what a bit index is, and its description of "3-bit unsigned bit index" clarifies its meaning.
• Yeah, I was having a hard time phrasing that line well. Happy to change it, particularly if we document STOP as STOP n8. (Or STOP [n8] since the parameter is optional, but then it looks like dereferencing a memory location...)

One important thing: The "bitwise NOT" was not meant for CCF but CPL. Because CCF affects a flag which is basically a boolean, CCF would be called the "logical NOT". I personally wouldn't even use a term here at all, but say that this "inverts" or "flips the carry flag", though I'm not sure if this wording makes sense to other people and where to put this.

You could also call CPL the "subtraction from 255" (A = 255-A).

CPL is the ComPLement (aka bitwise NOT) of register A. CCF is the Complement of the Carry Flag (again aka bitwise NOT, or logical negation in this case since it's one bit).

Damn the diff is impossible to read with just one word per line. How do manpage maintainers do this?

Note that using Split diff (click the little cog menu in the Files tab) is much more readable, at least for this kind of diff. (I sometimes switch back to Unified for some code changes, but rarely.)

I agree that “bitwise NOT” should be called out for CPL, not CCF. Since the latter only affects a single bit, calling it logical or bitwise is IMO just pedantry. CPL, however, has the concept being much more load-bearing.