In oxc_ast we exclusively use Box and Vec from oxc_allocator, so it's easy to know that Box or Vec anywhere in the code means the arena variant.

But this is not the case in transformer, where we are using both oxc_allocator::Vec and std::vec::Vec alongside each other. It's often confusing which Vec refers to, and consequently we've made mistakes, allocating some of transformer's temporary data structures into the arena by accident.

I suggest that we using naming to remove this ambiguity. We could either:

1. Rename the types

• oxc_allocator::Box becomes ABox
• oxc_allocator::Vec becomes AVec.

Now anywhere the code says Vec, it's std::vec::Vec, and anywhere it says AVec, it's an arena vec.

We could also use longer names ArenaBox and ArenaVec, but these types are very commonly used, so I think a shorter name is preferable.

2. Naming convention

Don't rename the types.

Instead, adopt a naming convention in transformer (and probably minifier too) when importing arena box/vec:

use oxc_allocator::Box as ABox;
use oxc_allocator::Vec as AVec;

This is less fool-proof - it's easy to forget a naming convention - but it's a less intrusive change.

Which?

I'm on the fence. Open to opinions.