hilo error when sampling distribution has missing values

[robjhyndman]

library(distributional)
library(magrittr)
dist_sample(list(c(NA,rnorm(20)))) %>% hilo()
#> Error in quantile.default(X[[i]], ...): missing values and NaN's not allowed if 'na.rm' is FALSE

^{Created on 2020-04-10 by the reprex package (v0.3.0)}

I think it would be reasonable to generate a warning here rather than an error, and pass na.rm=TRUE to quantile().

[mitchelloharawild]

Is there any value to having NA values in dist_sample()? Perhaps the warning should occur on object creation.

[robjhyndman]

Yes, that would be an alternative (or warn in both functions). There is no value in having NAs, but sometimes they might be generated. For example, if the distribution is a set of percentiles, then it is possible that some NAs will be generated due to the percentiles not covering the full range (0,100).

[mitchelloharawild]

Fixed, thanks.