Number of rows and list columns with NULL

[sluga]

The following snippet produces what it should, i.e. a list column axis with NULL values.

library(data.table) # v1.18.0

dt1 <- data.table(var_name = c("foo", "bar"))
dt1[, axis := lapply(var_name, \(v) NULL)]

But if the table happens to contain a single row, it attempts to remove the column instead, i.e. it behaves like dt2[, axis := NULL] would.

dt2 <- data.table(var_name = c("foo"))
dt2[, axis := lapply(var_name, \(v) NULL)]

[ben-schwen]

Even simpler example.

dt1 <- data.table(var_name = c("foo", "bar"))
dt1[, axis := list(NULL, NULL)]

dt2 <- data.table(var_name = c("foo"))
dt2[, axis := list(NULL)]

The problem is that when the rhs of := is list(NULL) it unwraps everything because it thinks, you have made an mistake. You can mitigate this by wrapping it into an extra list, e.g. with dt2[, axis := list(list(NULL))] or set(dt2, j="axis", value=list(list(NULL)))

edit: Apparently you can also use the functional form of := with dt2[, :=(axis=lapply(var_name, \(v) NULL))]