dplyr::cummean unexpected results

[reinholdsson]

I get unexpected results when using dplyr::cummean, see the following example:

a <- rep(99, 9)
a == dplyr::cummean(a)
# [1]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE FALSE FALSE
as.integer(dplyr::cummean(a))
# [1] 99 99 99 99 99 99 98 98 98

with base R:

a == base::cumsum(a)/1:length(a)
# [1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE

I'm using the latest version of dplyr from github (master branch).