Why is nfinit(f).zk not the same as nfbasis(f) ??

[Jeroen Demeyer <jdemeyer@cage.ugent.be>]

Hello pari-users,

With the svn version of PARI, I get:

gp> f = x^3 + x^2 - 2*x + 8
%1 = x^3 + x^2 - 2*x + 8
gp> nfbasis(f)
%2 = [1, x, 1/2*x^2 - 1/2*x]
gp> nfinit(f).zk
%3 = [1, 1/2*x^2 + 1/2*x - 1, x]

Is it normal to get different results here?  I agree that a basis is not
canonical, but it's not clear to me why nfbasis() uses a different
algorithm than nfinit().

Jeroen.