Using 'on' style join with two different column names

[peterlittlejohn]

Dear data.table team,

I came across a small issue with the use of 'on' in table joins. Or am I misunderstanding the documentation (pasted in below posting):

`

R version 3.3.2 (2016-10-31) -- "Sincere Pumpkin Patch"
Platform: x86_64-redhat-linux-gnu (64-bit)
library(data.table)
data.table 1.10.4

d1 <- data.table(A=c(1,2,4),B=c("a","b","d"))
d2 <- data.table(A=c(1,1,4))
d1[d2,on="A"]
A B
1: 1 a
2: 1 a
3: 4 d`

All works well. But now when I try a different column name in d2, it does not work:

`

d1 <- data.table(A=c(1,2,4),B=c("a","b","d"))
d2 <- data.table(W=c(1,1,4))
d1[d2,on=c(x="A",y="W")]
Error in [.data.table(d1, d2, on = c(x = "A", y = "W")) :
Column(s) [x,y] not found in x
`
Am I misunderstanding the documentation (?data.table), which states (in one bullet point on the use of "on"):

• As a named character vector, e.g., X[Y, on=c(x="a", y="b")]. This is useful when column names to join by are different between the two tables.

Given my understanding over this sentence in the documentation, the code above should produce the identical output as the first example.

[MichaelChirico]

Sorry, you've misunderstood the documentation. Maybe the wording needs to be cleaned up.

The names of the vector passed to on must be column names in x, and the values must be columns in i:

d1[d2, on = c(A = "W")]

[HughParsonage]

You can also do

d1[d2, on = "A==W"]