Re: Reducible thue() might be missing solutions

[Bill Allombert <Bill.Allombert@math.u-bordeaux.fr>]

On Sat, May 10, 2014 at 07:09:12PM +0200, Karim Belabas wrote:
> 1) Here's a minimal fix for 2.7.*
> 
> diff --git a/src/basemath/QX_factor.c b/src/basemath/QX_factor.c
> index 87190dc..2efb912 100644
> --- a/src/basemath/QX_factor.c
> +++ b/src/basemath/QX_factor.c
> @@ -807,9 +807,9 @@ DDF_roots(GEN A)
>    { lc = NULL; lcpol = A; }
>    else
>    { lc = absi(lc); lcpol = ZX_Z_mul(A, lc); }
> -  Ap = ZX_to_Flx(A, pp);
> +  Ap = Flx_normalize(ZX_to_Flx(A, pp), pp);
>    bound = root_bound(A);
> -  if (lc) { Ap = Flx_normalize(Ap, pp); bound = mulii(lc, bound); }
> +  if (lc) bound = mulii(lc, bound);
>    e = logint(addiu(shifti(bound, 1), 1), p, &pe);
>    pes2 = shifti(pe, -1);
>    if (DEBUGLEVEL>2) timer_printf(&T, "Root bound");

Which you applied to pari 2.7.1.

Thus, should I make a 3rd prerelease of pari-2.7.1 with this change ?

Cheers,
Bill.