Re: polredabs(,16)

[Igor Schein <igor@txc.com>]

On Wed, Sep 11, 2002 at 04:39:28AM +0200, Karim BELABAS wrote:
> On Tue, 10 Sep 2002, Igor Schein wrote:
> > here's my typical case:
> >
> > \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
> > ? \g1
> >    debug = 1
> > ? polredabs(...,16)
> [...]
> > The polynomial is not reduced, and the only way I know about it is if
> > I run at \g, otherwise it's completely silent.  I would like to have
> > an option to have
> > 62277548538789561520401660217885073427574453048708934544094318214969928190701341602751
> > from the example above factored:
> >
> > ? factor(62277548538789561520401660217885073427574453048708934544094318214969928190701341602751)
> >
> > [524351 15]
> >
> > So basically, leave polredabs(,16) behave as it does now, and have,
> > say, polredabs(,24) factor JUST the composites that appear in
> > impossible inverse.
> 
> Not necessary. It was a bug in allbase(), introduced by my recent patch [ try
> to recover when exception "impossible inverse mod..." is raised ]. It is
> allowed to have pseudoprimes in the discriminant factorization, but it is
> crucial that these be coprime !
> 
> I have modified the recovery code to enforce this (thereby discovering new
> factors, and reducing the number of failures). Does any of your examples
> break it ?

No, it's clean now.  Somehow, I didn't think of it as a bug.

Thanks

Igor