Inconsistent behaviour of rnfeltabstorel()

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

Hello pari-dev,

I'm not sure if this a bug or not, but at least it looks strange to me:

gp> K = nfinit(y^3-2);
gp> L = rnfinit(K, x^2 - Mod(y,K.pol));
gp> rnfeltabstorel(L, x^2)
%14 = Mod(y, x^2 - y)
gp> rnfeltabstorel(L, x^2+x)
%15 = Mod(Mod(1, y^3 - 2)*x + Mod(y, y^3 - 2), x^2 - y)

This looks inconsistent to me. Why is %14 equal to
Mod(y, x^2 - y)
instead of (what I would expect)
Mod(Mod(y, y^3 - 2), x^2 - y)

It also seems to conflict with ??rnfeltabstorel which says that the
result is a polmod with polmod coefficients.


Cheers,
Jeroen.



PS for Karim and Bill: adding you as cc because of recent trouble with
the pari-dev list.