Is it logical or acceptable to approximate a distribution of a random variable X of infinite mixture gaussian distributions to a uniform distribution [a,b]?

where :

1- Each gaussian distribution has a mean muii where a < muii < b and,

2- muii = muii-1 + epsilon: where epsilon --> 0

2- Variance segma is constant.