Created NULL column somehow?

[MichaelChirico]

Was trying to swap values from a column into another column and erase the original values; I made a mistake in doing so:

set.seed(20349)
DT = data.table(
  a = rnorm(10)
)
DT[a < 0, c('a', 'b') := .(NULL, a)]

Which correctly errors:

Error in [.data.table(DT, a < 0, :=(c("a", "b"), .(NULL, a))) : When deleting columns, i should not be provided

However, now DT is corrupted:

DT
#              a    b
#  1:  0.6667242 NULL
#  2: -1.8261938 NULL
#  3:  0.2047286 NULL
#  4: -0.4425548 NULL
#  5: -0.6238520 NULL
#  6: -1.9680856 NULL
#  7:  0.2002663 NULL
#  8: -0.1607745 NULL
#  9:  0.0206196 NULL
# 10:  0.3074775 NULL

DT$b
# NULL

DT[is.na(b)]

Error in [.data.table(DT, is.na(b)) : SETLENGTH() can only be applied to a standard vector, not a 'NULL'

DT[ , b := NULL]

Error in ans[[target]] : subscript out of bounds

Seems the only way to clean the table is with a copy:

DT = DT[ , !'b']

Obviously what I tried is wrong (I meant to have .(NA, a) on the RHS), but it seems this mistake has cascaded...

[DavidArenburg]

You can also remove it using DT[["b"]] <- NULL (which also makes a copy). Also DT[1] crashes my R session. Interesting edge case

[mrdwab]

In case it helps, set seems to work:

packageVersion("data.table")
# [1] ‘1.11.8’
set(DT, j = "b", value = NULL)
str(DT)
# Classes ‘data.table’ and 'data.frame':	10 obs. of  1 variable:
#  $ a: num  0.667 -1.826 0.205 -0.443 -0.624 ...
#  - attr(*, ".internal.selfref")=<externalptr>