Dynamically operate on column (inside :=)

Question

I have the below data table

library(data.table)

df <- data.table(col1 = 1:3
                 , col2 = 2:4
                 )

I would like to dynamically operate on a column, whose column name is stored in the below object:

metric <- 'col2'

but the below failed

df[, `:=` (col1 = col1 - 1
           , as.name(metric) = as.numeric(get(metric))
           )
   ]

error:

Error: unexpected '=' in:
"df[, `:=` (col1 = col1 - 1
           , as.name(metric) ="

I would like to operate on col1 and whatever column whose name is stored in metric in one call. Are there any methods other than the below (which works)

df[, c('col1', metric) := .(col1 - 1
                            , as.numeric(get(metric))
                            )
   ]

Answer 1

Using get() in data.table has been retired. I think the proper way to do this these days is using env.

I'll add 2 to the assignment so we can see it is modifying something:

df[, `:=`(
    col1 = col1 - 1,
    col = as.numeric(col) + 2
), env = list(col = metric)][]

#     col1  col2
#    <num> <num>
# 1:     0     4
# 2:     1     5
# 3:     2     6

Or using let(), which is an alias for `:=` as suggested by Roland in the comments:

df[,
    let(
        col1 = col1 - 1,
        col = as.numeric(col) + 2
    ),
    env = list(col = metric)
] # same result