r - How to select columns programmatically in a data.table? -

i have following data.table (dt):

dt <- data.table(v1 = 1:3, v2 = 4:6, v3 = 7:9) 

i select subset of variables programmatically (dynamically), using object relevant variable names stored. example, want select 2 columns "v1" , "v3" stored in variable "keep"

keep <- c("v1", "v3") 

if select "keep" columns data.frame, following work:

dt[keep] 

unfortunately, not working when data.table. thought data.frame , data.table identical kind of behavior, apperently aren't. able advise on correct syntax?

this covered in faq 1.1, 1.2 , 2.17.

some possibilities:

dt[, c('v1','v3'), = false] dt[, c(1,3), = false] dt[, list(v1,v3)] 

the reason df[c('v1','v3')] works data.frame covered in ?`[.data.frame`

data frames can indexed in several modes. when [ , [[ used single vector index (x[i] or x[[i]]), index data frame if list. in usage drop argument ignored, warning.

---

from data.table 1.10.2, may use .. prefix when subsetting columns programmatically:

when j symbol prefixed .. looked in calling scope , value taken column names or numbers [...] experimental.

thus:

dt[ , ..keep] #    v1 v3 # 1:  1  7 # 2:  2  8 # 3:  3  9