Better selection of name for ReassignmentToVal error message

[jan-pieter]

(from today's Scala 3 compiler spree)

Use name from tree when it is a NameTree for the ReassignmentToVal error message

Fixes #22671

[SethTisue]

assigning @mbovel as punishment for deserting us :-P

I put some commentary on the ticket about how we found our way to this approach

so then our questions for review are:

1. is NameTree even the way to go here?
2. can we assume that lhsCore is always a NameTree? does it make sense to keep the fallback case _ => lhsCore.symbol.name case?

[SethTisue]

I'm also wondering why it wasn't done this way in the first place, tbh. like does using a symbol (instead of a plain name) in an error message sometimes improve message quality somehow...?

[som-snytt]

If I import a symbol, it might be renamed. Also, LHS of an assignment can be map(k) style.

Scala 2 expends some LOC on this error case. There is advice1 for does not convert to assignment and advice2 for receiver is not assignable.

[SethTisue]

I see. In that case, it sounds like it's worth doing some further exploring of whether we can get a hold of the right val with a little more work.

[SethTisue]

(I've assigned myself as a backstop, but @iusildra @jan-pieter @ajafri2001 if any of you wants to dig into this first, feel free, as I won't get to it right away.)

[som-snytt]

I submitted #22869 where I'll add spree participants and deserters as co-conspirators. That just uses the other tree. But I'd also like to add the addendum when += fails to become update because of errors; that is fun syntax that is possibly underused.

[som-snytt]

Many months ago, I saved the format of the co-author tag so I wouldn't forget it, and what I saved was:

➜  ~ cat co-author

    Co-authored-by: Jan-Pieter van den Heuvel
    <[email protected]>