Lag behaves differently inside and outside of mutate

[matfournier]

From:

http://stackoverflow.com/questions/28038370/lag-changed-in-dplyr-0-4-1-default-parameter-not-working-inside-of-mutate

Trivial example that worked in previous dplyr versions but not 0.4.1, as the first row is left as NA rather than the default value:

blah <- data.frame(x1 = c(5,10,20,27,35,58,5,6))
blah <- mutate(blah, x2 = x1 - lag(x1, n=1, default=0))

Will return NA on the first row. Oddly lag(x1, default=0) works as intended.

Doing the above code outside of the mutate function works fine (accepts default=x param):
e.g. blah$x2 <- blah$x1 - lag(blah$x1, n=1, default=0) works as intended.

Thus lag works differently inside and outside of mutate.

[hadley]

Not a crasher, so putting off until 0.5.

[tentacles-from-outer-space]

Just to notice. Same thing happens with order_by. Following example from help("lead"):

right <- mutate(scrambled, prev = lag(value, order_by = year))
arrange(right, year)

doesn't work as expected when n is added:

right <- mutate(scrambled, prev = lag(value, n=1L, order_by = year))
arrange(right, year)