Append a "select all" DT[] to translation if the last verb is mutate #146

data.table's := syntax has a side-effect of messing with printing (see Bug Fixes in data.table 1.9.6):

suppressPackageStartupMessages(library(data.table))

z <- data.table(a = 1)[, b := 1]
z
z
#>    a b
#> 1: 1 1

^{Created on 2020-01-29 by the reprex package (v0.3.0)}

Note how the first time we ask to print z, nothing happens. Only after the second print does anything happen.

The solution is to add a final [] to the query:

suppressPackageStartupMessages(library(data.table))

z <- data.table(a = 1)[, b := 1][]
z
#>    a b
#> 1: 1 1

This behavior is also present in dtplyr whenever a mutate is the last verb:

suppressPackageStartupMessages({
  library(data.table)
  library(dplyr)
  library(dtplyr)
})

z <- data.table(a = 1) %>%
  lazy_dt() %>%
  mutate(b = 1) %>%
  as.data.table()
z
z
#>    a b
#> 1: 1 1

An obvious workaround is to add a select(everything()) after the mutate.

While this is arguably best resolved from the data.table side, it isn't consistent with dplyr, where such a query should print from the get-go. In order to facilitate potential migration from dplyr to dtplyr, it'd be nice to solve this here.

I'm assuming the performance impact of DT[] is negligible, so a trivial solution would be to append a [] at the end of all translations, though a more elegant solution (if plausible?) would only do so when the last verb is a mutate.