+ Nick and Guido

The only benefit I see on our side is that it leaves the door open for complex assignment targets in the future, like (a, b), a[b], etc.

(If I recall correctly, this is also why NamedExpr uses an expr target rather than just an identifier.)

I guess I'm neutral on this. The usability argument seems weak, but I can see the separate case for forward-compatibility with possible syntax extensions in the future. Thoughts?

Marking as high priority since we need to make a decision on this before the beta.