Fix #1356

[devmotion]

Fixes #1356.

[codecov-commenter]

Codecov Report

Merging #1357 (61fed19) into master (a6b9960) will increase coverage by 0.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1357      +/-   ##
==========================================
+ Coverage   82.74%   82.76%   +0.01%     
==========================================
  Files         116      116              
  Lines        6672     6672              
==========================================
+ Hits         5521     5522       +1     
+ Misses       1151     1150       -1     

Impacted Files	Coverage Δ
src/univariate/discrete/betabinomial.jl	88.46% <ø> (+1.92%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a6b9960...61fed19. Read the comment docs.

[mschauer]

What about cdf(d::BetaBinomial, k::Real)?

[devmotion]

There's no need to implement it, the default fallback in

Distributions.jl/src/univariates.jl

Line 351 in a6b9960

cdf(d::DiscreteUnivariateDistribution, x::Real) = cdf_int(d, x)

is sufficient. In general, for discrete distributions whose support is of type Int one just has to define cdf(d, ::Int) (otherwise one has to implement cdf(d, ::Real)). One does not have to define ccdf, logcdf, logccdf (they are computed from cdf) but it is numerically more stable to compute them by direct summation (with logsumexp for the log variants).