filter by variable bug

[thomasjwood]

When we filter a table by a variable, unexpected behavior results. Imagine this tbl

d1 <- tbl_df(data.frame(num1 = as.character(sample(1:10, 1000, T)),
                        var1 = runif(1000)))
d1$num1 <- as.character(d1$num1)

which I want to subset according to values I find in data frame d2

d2 <- data.frame(num1 = 1:3)
d2$num1 <- as.character(d2$num1) 

If I pass the num1 values as a separate character vector, everything works great. For instance, this

d1 %.%
  filter(num1 %in% c("1", "2", "3")) %.%
  group_by(num1) %.%
  summarise(mu = mean(var1))

returns

Source: local data frame [3 x 2]

  num1        mu
1    2 0.4662081
2    1 0.4810027
3    3 0.4920704

But when filtering with a variable:

d1 %.%
   filter(num1 %in% d2$num1) %.%
   group_by(num1) %.%
   summarise(mu = mean(var1))

dplyr returns this in version 0.1.1

Source: local data frame [0 x 2]

and this in version 0.1.2

Error in list(num1 = c("1", "2", "3"))$c(7L, 2L, 3L, 4L, 10L, 6L, 7L,  : 
  invalid subscript type 'integer'