Is there a reason why all discrete variables lack the logcdf method? Stan, for example, provides this functionality for several distributions (Poisson, Binomial, Negative binomial, Bernoulli) Is this something people might be interested? Best, Ricardo