Can you confirm that this is a pure refactoring that doesn't change any APIs or semantics? (That's what it sounds like, but there's a lot of code in the diff...)

Yeah, it is almost exclusively only moving code around in the file. The only 2 exceptions are the addition of decode_linetable() and new_linesiterator(), which are just code_getlnotab() and code_linesiterator() (respectively) with tighter signatures.

Looks like you broke the argument clinic?

This is almost certain to cause merge conflicts with work on PEP 657.
We will be removing all the opcache stuff in a month or two (PEP 659) so I wouldn't bother moving that around.

Given that the line number table is likely to grow and change with PEP 657, how about moving that code to it's own file?
@pablogsal thoughts?

Given that the line number table is likely to grow and change with PEP 657, how about moving that code to it's own file?
@pablogsal thoughts?

I think having it in the same file is fine, as it will cause merge conflicts for us in any case. There is some benefit of having fewer compilation units. In the other hand maybe it will result in a better organisation of the code so if you think it makes sense I would be fine too :)

Let's not sit on this longer. LGTM, if either Mark or Pablo approves then just land it.