Re: bug: (1+O(x^2))/x -> ** not enough memory

[Karim BELABAS <Karim.Belabas@math.u-psud.fr>]

[Xavier:]
> with pari-2.0.20 (beta)
> 
> ? (1+O(x^2))/x 
> %1 = x^-1 + O(x)
> 
> with the CVS version
> 
> ? (1+O(x^2))/x 
>   ***   not enough memory
> 
> Looks like it's a 64bits only bug. Bug starts with line 1876 of gen1.c.
> At this point, PARI thinks lgef(x) is 2147483648, so the next command 
> 
>  y = (GEN) gpmalloc(l*sizeof(long));
> 
> line 330 of gen2.c (with l being lgef(x)) exhaust all the available
> memory.

It's a very recently introduced bug, meant to correct another one (related
to multivariate power series). I thought there was a typo here and that
lgef really meant lg, but it didn't (mistook x for y, the power series for
the polynomial...).

It's corrected.

Sorry,

    Karim.
-- 
Karim Belabas                    email: Karim.Belabas@math.u-psud.fr
Dep. de Mathematiques, Bat. 425
Universite Paris-Sud             Tel: (00 33) 1 69 15 57 48
F-91405 Orsay (France)           Fax: (00 33) 1 69 15 60 19
--
PARI/GP Home Page: http://www.parigp-home.de/