Split parser into main entrypoint and two helpers

[neuroevolutus]

parse now calls into parse_expression and is a bit easier to read.

[tekknolagi]

Ooh, I like that this cleans up the massive function. I wonder about the names. Both functions are parsing expressions and I don't know how name a meaningful difference between the two. The latter feels more cleanly like a parse_binop. I would be interested in hearing your thoughts on this. Maybe parse_unary? parse_atom?

[tekknolagi]

Unrelated, I gotta figure out how to turn off preview deploys and building Docker images for others' PRs. Sorry about the red Xes.

[neuroevolutus]

Unrelated, I gotta figure out how to turn off preview deploys and building Docker images for others' PRs. Sorry about the red Xes.

No worries!

[neuroevolutus]

Ooh, I like that this cleans up the massive function. I wonder about the names. Both functions are parsing expressions and I don't know how name a meaningful difference between the two. The latter feels more cleanly like a parse_binop. I would be interested in hearing your thoughts on this. Maybe parse_unary? parse_atom?

I think that parse_unary would be a fine name for the new function. parse_binop would also be a great choice for renaming the existing parse function. Alternatively, since parse is responsible for parsing an entire program (as far as I understand), what would you think of renaming it to parse_program?

As a side note, I really appreciate how you have a lot of unit tests in the code. It really does make a world of a difference when trying out small refactors such as this.

[tekknolagi]

Let's go with parse_unary and parse_binary and then we can do parse = parse_binary. Alternately, we can make parse an entirely new function that doesn't take a precedence and then make the precedence non-optional for the other two internal functions. I leave this decision up to you; both are fine with me. After these renames, I would be delighted to merge.

[tekknolagi]

Thank you!