long time bug

[Igor Schein <igor@txc.com>]

Hi,

I came across this behavior, going back at least as far
as 2.0.11:

? bnf=bnfinit(polcyclo(5,y));
? rnfconductor(bnf,x^2-2);
? nfsubfields(bnf);
? rnfconductor(bnf,x^2-2);
  ***   incorrect polynomial in rnf function.

As you can see, rnfconductor() works fine the first time
around, but then ( I guess ) something gets corrupted
during nfsubfields(), and it fails the 2nd time.

Thanks

Igor