Re: PATCH: polcyclo_eval in roots of unity

[Bill Allombert <Bill.Allombert@math.u-bordeaux1.fr>]

On Tue, Feb 10, 2009 at 12:46:18PM +0100, Jeroen Demeyer wrote:
> Hello pari-dev,
> 
> This patch fixes the libpari function polcyclo_eval (the GP function 
> polcyclo) such that evaluating in a root of unity now works. 
> Additionally, I changed it such that it can compute n-th cyclotomic 
> polynomials where n is a t_INT (instead of a long).

I am concerned that polcyclo_eval is going the route of substpol:
marginally useful, very complex to implement properly and never
fully tested. Allowing n to be a t_INT is another increase in
complexity. 

On top of that, the current polcyclo under GP remove an useful level
of error checking: polcyclo(n, y) used to report an error when y was
already used, which is no more the case. 

Cheers,
Bill.