Unable to use eval(parse(...)) in data.table [,:=] function

Question

I am not sure if I made an error here (dangers of using eval(parse(...))perhaps), but here I tried to use eval(parse(...)) while updating a data.table column, which creates an error. Below is my code.

require(data.table)
# Loading required package: data.table
# data.table 1.9.2  For help type: help("data.table")
DT<-data.table(x=seq(1,10,1),
               y=seq(2,20,2))
DT[,y:=y*eval(parse(text="1*2"))] #Doesn't work
# Error in y * expression(1 * 2) : non-numeric argument to binary operator
variable <- eval(parse(text="1*2"))
DT[,y:=y*variable] #Works
DT
#     x  y
# 1:  1  4
# 2:  2  8
# 3:  3 12
# 4:  4 16
# 5:  5 20
# 6:  6 24
# 7:  7 28
# 8:  8 32
# 9:  9 36
#10: 10 40

Answer 1

This behaviour, bug #5423, is now fixed in commit #1234 in the current development version of data.table v1.9.3. From NEWS:

o DT[,y := y * eval(parse(text="1*2"))] resulted in error unless eval() is wrapped with paranthesis. That is, DT[,y := y * (eval(parse(text="1*2")))]. Now fixed, closes #5423. Thanks to Wet Feet for reporting and to Simon O'Hanlon for identifying the issue here on SO.

---

Regardless of why you want to do this and the relative merits or not, this is about order of precedence. You need to wrap the whole eval(parse(text="1*2")) in a set of parentheses to ensure that it is evaluated before you try to multiply it by y...

DT[,y:=y*(eval(parse(text="1*2")))] 
#     x  y
# 1:  1  4
# 2:  2  8
# 3:  3 12
# 4:  4 16
# 5:  5 20
# 6:  6 24
# 7:  7 28
# 8:  8 32
# 9:  9 36
#10: 10 40

Without the parentheses data.table::":=" is trying to multiply y by an expression object, which will result in a syntax error.