Make dcast.data.table work with undefined variables

[DavidArenburg]

library(data.table) ; library(reshape2) 
dcast(df, cumsum(V1 == 0) ~ V1, value.var = 'V2') # Works
#   cumsum(V1 == 0)     0     1     2     3      4
#1               1 1.052 0.542 0.496 0.402 0.2780
#2               2 5.115 4.329 4.121 4.075 4.0088
dcast(setDT(df), cumsum(V1 == 0) ~ V1, value.var = 'V2') # Doesn't work
# Error in dcast.data.table(setDT(df), cumsum(V1 == 0) ~ V1, value.var = "V2") : 
#   Column 'cumsum(V1==0)' not found.

Data set

df <- structure(list(V1 = c(0L, 1L, 2L, 3L, 4L, 0L, 1L, 2L, 3L, 4L), 
          V2 = c(1.052, 0.542, 0.496, 0.402, 0.278, 5.115, 4.329, 4.121, 
          4.075, 4.0088)), .Names = c("V1", "V2"), class = "data.frame", row.names = c(NA, -10L))

[arunsrinivasan]

The next version of dcast will be able to do this.

I did not know this was even possible with reshape2::dcast. I don't think there are examples showcasing this feature.

[mrdwab]

@arunsrinivasan There was this. :-)