Christian Maeder schrieb: [...]
Without this restriction polynomials like "- a + - b * x + - c * - x ^ 2" would uniquely resolve to "((- a) + (- (b * x))) + (- (c * (- (x ^ 2))))"
I think hugs handles this correctly!
yes it does this case.
Let us assume a user-defined (non- or) right-associative operator "#" with the same precedence as "+" and "-" (infix[r] 6 #).
3.c) both "- a # b" and "a # - b" are rejected, because "#" is not left-associative (like "-").
hugs rejects only the case "- a # b": ERROR - Ambiguous use of unary minus with "(#)" So hugs still considers associativity for "-" (but does not consider the operator to the left of an unary minus.) C.
This unnecessary restriction rules out a (user-defined) polynomial like "- a # - b * x"