Addition of expression and order term fails

[avichaldayal]

a = (x/2)**n*(1/(factorial(n)) - x**2/(4*factorial(n + 1)) + x**4/(32*factorial(n + 2)) + O(x**6))

o = O(x**6)

a + o fails.

Related to #7552

[skirpichev]

simple test: O(x**n, x) + O(x**2)

[avichaldayal]

Some of the special functions like bessel functions take two parameters - n and x.
In those cases, when finding series around x, perhaps it should be assumed that n is very small? In that case, the x**n terms won't be removed by the order term.
But I'm not sure if that kind of assumption exists and if its even the right approach

[skirpichev]

On Thu, Jun 19, 2014 at 10:39:44AM -0700, Avichal Dayal wrote:

Some of the special functions like bessel functions take two parameters -
n and x.
In those cases, when finding series around x, perhaps it should be assumed
that n is very small?

Why? _eval_nseries (if you are talking about) - is for support of
univariate series. There is nothing specific about arguments of
function, except that they are functions of x (I don't sure we can
handle the case when they both are function of x).