Built on top of Distributions.jl, the goal of MixedDistributions is to represent and manipulate discrete-continuous variables.
An informal definition of such variable is a continuous distribution with mass points (the CDF has positive discontinuities at these mass points).
https://github.com/matbesancon/MixedDistributions.jl
As I see it for now, this could look like mixture distributions, but formulated in a more general way than what Distributions.jl develops for its AbstractMixtureModel type which is parametrized on one family of distributions.