rdunif unexpected output

[1danjordan]

When rdunif has an upper bound b less than the lower bound a, it outputs b n times.

rdunif(10, -5, 5)
# [1] -5 -5 -5 -5 -5 -5 -5 -5 -5 -5

A solution is to pass a vector of integers into sample

rdunif2 <- function(n, b, a = 1) {
  vec <- a:b
  sample(vec, n, replace = TRUE)
}

rdunif2(10, -5, 5)
# [1] -2  3  2 -4 -4  3  3  2  4 -5

I've benchmarked them and rdunif2 is slightly slower.

microbenchmark(
   rdunif(1000, 50),
   rdunif2(1000, 50)
)
# Unit: microseconds
#             expr    min      lq     mean median      uq     max neval cld
#  rdunif(1000, 50) 23.179 29.8285 35.06438 31.919 34.5780 125.010   100  a 
# rdunif2(1000, 50) 30.019 36.0980 42.52322 37.997 39.8975 139.069   100  b

But maybe acceptably so?

[hadley]

Passing a vector doesn't work in general because sample(10:10) probably doesn't do what you expect.

[1danjordan]

Ah I missed that! You don't need my suggestion, but I then I suppose this would work

rdunif2 <- function(n, b, a = 1) {
  if(a == b) return(rep(a, n))
  vec <- a:b
  sample(vec, n, replace = TRUE)
}

Can make a pull req if you want. Apologies if I'm distracting you from the important stuff.