In 30741 I am trying to address some issues with PurePoly and am running into some API questions.
I think a coherent way to use M.charpoly is like
M.charpoly() # PurePoly
M.charpoly(x) # Poly <-- (*)
M.charpoly().as_poly(x) # Poly
M.charpoly()(x) # Expr
___________________
(*) in master charpoly(x) creates a PurePoly with symbol 'x' for display purposes
In the PR, generators are canonical so hashing and equality work as they should. e.g. if `p = PurePoly(x)` and `q = PurePoly(y)` then `hash(p)==hash(q)` and `p==q`.
But master allows `PurePoly(x) == Poly(x)` which seems wrong. The consequence is that if we disallow that equality (like we don't allow `Integer(1)==Float(1)`) then letting `charpoly(x)` produce a `Poly` would break the current contract that `charpoly(x) == charpoly(y)`.
How significant would that sort of change be?
/c