Sorry for taking a while to review this. I started adding lots of detailed comments, but have abandoned that for now, because my overall feeedback took shape at a higher level -- I don't think lexical analysis is the right place for large parts of this content. For example, explaning the detail of how replacement fields work necessarily references expressions, which we haven't yet covered in the PLR, as we're still notionally in the tokenising/lexing phase.

From the introduction of the chapter, 'This chapter describes how the lexical analyzer breaks a file into tokens.', and I think we make a mistake by trying to cover/explain all of f-strings here.

I wonder if there is a clear enough distinction we can draw between the lexing of f-strings (that could go into lexical_analysis, the parsing & grammar thereof (that could go into e.g. expressions(?)), and the runtime evaluation of the AST, e.g. how format specs are evaluated (that could go into stdtypes?)

I think there are bits of the treatment & grouping that are somewhat nebulous and could go either way, but I'd prefer to be stricter with the content in the Language Reference, even if that means splitting up the coverage of f-strings into a few places.

I'd be happy to work up a PR with something closer to my suggested split if it'd be useful, but otherwise let me know your thoughts.

A