fix(compiler): Fix array infix assign semantics

[ospencer]

Closes #1988

[spotandjake]

This looks great, as a note my pr that implemented infix formatting still serves a role as in cases like:

arr[1] = arr[1] + 1

We are still going to want the formatter to convert those to infix automatically.

[ospencer]

I didn't implement that because of the arr[foo()] case. I'm not sure if we want to do that change when it's semantically correct, because it's pretty annoying to detect.

[spotandjake]

I didn't implement that because of the arr[foo()] case. I'm not sure if we want to do that change when it's semantically correct, because it's pretty annoying to detect.

Detecting the simple case of an identifier or constant is easy I just don't think we should go beyond that .

[ospencer]

I'll do that in a follow-on PR.

[spotandjake]

Sounds good

[ospencer]

I'll also do a follow-on PR to use this same format for regular infix assignment and record infix assignment.

[alex-snezhko]

Shouldn't this be an Impossible?

[ospencer]

It's not, because you can redefine what the infix operators do, and we don't stop you from making one take a number of arguments other than two. (You wouldn't be able to call it with the infix syntax, but you could call it as a regular function.)

[marcusroberts]

LGTM

[peblair]

Already merged, but also looks good to me