Is there any particular reason why computing the kernel of a matrix over a polynomial ring might fail? Consider the following example:

```
sage: PR.<x1, x2, x3, y1, y2, y3> = QQ[]
sage: def P(t):
... u = 1 - t
... x = x1*t*t + x2*t*u*2 + x3*u*u
... y = y1*t*t + y2*t*u*2 + y3*u*u
... return (x, y)
sage: def C(t):
... x, y = P(t)
... return vector(PR, [x^2, y^2, x*y, x, y, 1])
sage: M = Matrix([C(t) for t in [0, 1, -1, 2, 1/2]])
sage: M.right_kernel()
Traceback (most recent call last):
...
ArithmeticError: Ideal Ideal (x3^2, x1^2) of Multivariate Polynomial Ring
in x1, x2, x3, y1, y2, y3 over Rational Field not principal
```

It seems to me as if being able to compute the reduced echelon form (using the `frac`

method) should be enough to actually compute a kernel as well, without any additional requirements. I will demonstrate this in an answer below. So why does computing the echelon form work, but computing the kernel not?