Column type overwrite fails for recycled values

[MichaelChirico]

data.table(a = 1:2)[ , a := 'b']

Warning messages: 1: In [.data.table(data.table(a = 1:2), , :=(a, "b")) : NAs introduced by coercion 2: In [.data.table(data.table(a = 1:2), , :=(a, "b")) : Coerced 'character' RHS to 'integer' to match the column's type. Either change the target column ['a'] to 'character' first (by creating a new 'character' vector length 2 (nrows of entire table) and assign that; i.e. 'replace' column), or coerce RHS to 'integer' (e.g. 1L, NA[real|integer], as.*, etc) to make your intent clear and for speed. Or, set the column type correctly up front when you create the table and stick to it, please.

Not sure if this is intended behavior... it's technically against the rule specified by the warning, but seems like a natural usage

[MichaelChirico]

The problem is a bit more general -- length-1 attempts to plonk are first attempted to coerce to the original type:

data.table(a = rep(TRUE, 2L))[ , a := 0L][]         # a is now FALSE
data.table(a = rep(TRUE, 2L))[ , a := 0][]          # a is now FALSE
data.table(a = rep(TRUE, 2L))[ , a := 0+1i][]       # error (can't coerce complex->logical)
data.table(a = rep(TRUE, 2L))[ , a := "b"][]        # error, like in main issue
data.table(a = rep(TRUE, 2L))[ , a := .(list(0))][] # works with warning, probably not for other list inputs

data.table(a = 1:2)[ , a := TRUE][] # a is 1L,1L
data.table(a = 1:2)[ , a := 1][]    # a is 1L,1L
data.table(a = 1:2)[ , a := 0+1i][] # ...
data.table(a = 1:2)[ , a := "b"][]
data.table(a = 1:2)[ , a := .(list(1))][]

data.table(a = c(1,2))[ , a := TRUE][]
data.table(a = c(1,2))[ , a := 1L][]
data.table(a = c(1,2))[ , a := 0+1i][]
data.table(a = c(1,2))[ , a := "b"][]
data.table(a = c(1,2))[ , a := .(list(1))][]

data.table(a = rep(0+1i, 2L))[ , a := TRUE][]
data.table(a = rep(0+1i, 2L))[ , a := 1L][]
data.table(a =rep( 0+1i, 2L))[ , a := 1][]
data.table(a =rep( 0+1i, 2L))[ , a := "b"][]
data.table(a =rep( 0+1i, 2L))[ , a := .(list(1))][]

data.table(a = rep("b", 2L))[, a := TRUE][]
data.table(a = rep("b", 2L))[, a := 1L][]
data.table(a = rep("b", 2L))[, a := 1][]
data.table(a = rep("b", 2L))[, a := 0+1i][]
data.table(a = rep("b", 2L))[, a := .(list(1))][]

data.table(a = list(1, 2))[ , a := TRUE][]
data.table(a = list(1, 2))[ , a := 1L][]
data.table(a = list(1, 2))[ , a := 1][]
data.table(a = list(1, 2))[ , a := "b"][]
data.table(a = list(1, 2))[ , a := 0+1i][]