What causes a 'longer object length is not a multiple of shorter object length' in this snippet?

Question

Try the following R code (with tidyverse loaded):

dice = data.frame(roll = sample(x = 1:6, size = 1000, replace = TRUE))
ones = dice %>% filter(roll == 1)
length(ones$roll)

mutate(
    data.frame(n = 1:6), 
    len = length(filter(dice, roll == n)$roll)) 

The first 3 lines behave as expected, and give a sensible count for the number of ones rolled. The last line gives an error

1 1 152
2 2 152
3 3 152
4 4 152
5 5 152
6 6 152
Warning message:
In roll == n :
  longer object length is not a multiple of shorter object length

What am I doing wrong? Is it picking up n as a vector instead of operating on individual values?

NB. I know this is not a sensible way to count the no. elements with each value. It's just a convenient problem to illustrate the issue, which occurs in a much messier example.

Thanks!

Answer 1

dplyr thinks columnwise not rowwise. That mean it doesn't evaluate n = c(1,2,3,4,5,6) element after element, but all elements at once.

Doing:

mutate(
  data.frame(n = 1:6), 
  len = length(filter(dice, roll == n)$roll)) 

I get

  n len
1 1 164
2 2 164
3 3 164
4 4 164
5 5 164
6 6 164
Warning message:
In roll == n :
  longer object length is not a multiple of shorter object length

Which is the same as:

sum(dice$roll == 1:6)

[1] 164
Warning message:
In dice$roll == 1:6 :
  longer object length is not a multiple of shorter object length

Which compares two vectors at their position, recycling the shorter vector as often as necessary, giving a warning when the lenghts don't match.

If you put a rowwise() in between it evaluates n element after element:

data.frame(n = 1:6) %>% rowwise() %>% mutate(len = length(filter(dice, roll == n)$roll)) 

# A tibble: 6 x 2
      n   len
  <int> <int>
1     1   172
2     2   159
3     3   176
4     4   168
5     5   174
6     6   151