Mutating over a chain of two new columns works as expected:

memdb_frame(a = 1) %>% mutate(a2 = a, a3 = a2)

# Source:   lazy query [?? x 3]
# Database: sqlite 3.19.3 [:memory:]
      a    a2    a3
  <dbl> <dbl> <dbl>
1     1     1     1

However, three or more does not work for sql backends:

memdb_frame(a = 1) %>% mutate(a2 = a, a3 = a2, a4 = a3)

Error in rsqlite_send_query(conn@ptr, statement) : no such column: a3

This one is particularly important for sparklyr since this was supported in sparklyr 0.5 with a custom override and a user already contacted me pointing out their code is broken after upgrading to sparklyr 0.6 and dplyr 0.7, the workaround is to split as in:

memdb_frame(a = 1) %>% mutate(a2 = a, a3 = a2) %>% mutate (a4 = a3)

Related to #2483