It would be really nice if we could document our code in line. Right? I mean, look back at the metacircular evaluator. It had comments that made the functions so much clearer… but we couldn’t actually use that code because our interpreter has no idea what a semicolon is. Turns out this is really easy to add to our reader, so this post will be fairly short.
All we really need to do to add line comments is add the ability to say “if we see a semicolon, ignore the rest of the line”. That means we can do things like this:
$ ocaml 14_comments.ml > ((lambda ; this is a lambda (x) (+ x 1)) 4) 5 > Exception: End_of_file. $
While that comment is not at all useful, it does get ignored when evaluating the expression, which is what we want. Buf enough fluffing about. The implementation is a grand total of 4 lines:
let rec read_sexp stm = [...] let rec eat_comment stm = (* NEW *) if (read_char stm) = '\n' then () else eat_comment stm in eat_whitespace stm; let c = read_char stm in if c = ';' then (eat_comment stm; read_sexp stm) (* NEW *) else if is_symstartchar c [...]
And the code reads pretty much the same as the English. If we wanted to, we could pre-emptively eat comments like we do whitespace, or we could have a separate preprocessing step entirely — but this is fine for now.
Since we now support comments, I’ve transferred the comments from the
metacircular evaluator post into the new-and-improved version with
The results can be found here (lisp).
Download the code here (ml) if you want to mess with it.
Next up, printing. How else are we going to debug our code? /s