filter DT using max() in i: group_id==max(group_id)

[jangorecki]

Use case to pick only observations related to last group
Take following dt:

dt <- data.table(group_id = c(1,1,1,2,2,2,3,3,3), val = rnorm(9))

This works in 1.9.3 but doesn't work in 1.9.4:

dt[group_id == max(group_id)]

As for now to achieve the following in 1.9.4 I need to do:

dt[group_id == max(dt$group_id)]
# or
dt[,.SD][,max_group_id:=max(group_id)][group_id == max_group_id][,max_group_id:=NULL][]
# [,.SD] - prevent write dt by reference

Is there any recommended (in terms of future support) way of achieve that?

[arunsrinivasan]

Thanks @jangorecki. For now you can use options(datatable.auto.index=FALSE) to turn the feature off, and things should work normally, until we fix this one.