Memory efficient join

[matthewgson]

data.table has a great feature to join two large datasets efficiently by reference.
I'm wondering if this could be available in dtplyr.

Below are very nice details about the difference from StackOverflow (by Jaap)

library(bench)
bm <- mark(AA <- BB[AA, on = .(aa)],
           AA[BB, on = .(aa), cc := cc],
           iterations = 1)

> bm[,c(1,3,5)]
# A tibble: 2 x 3
  expression                         median mem_alloc
  <bch:expr>                       <bch:tm> <bch:byt>
1 AA <- BB[AA, on = .(aa)]            4.98s     4.1GB
2 AA[BB, on = .(aa), `:=`(cc, cc)] 560.88ms   384.6MB