Input data.table can be modified despite having immutable=TRUE in lazy_dt()

[jatherrien]

Despite specifying immutable=TRUE in lazy_dt, if I have a mutate followed by a summarize followed by a mutate, the first mutate will get applied to the input data.table object, when it should not be modified. I can't seem to trigger it unless I have a mutate after the summarize.

library(data.table)
library(dtplyr)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:data.table':
#> 
#>     between, first, last
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

d1= data.table(x=c(1,1,1,2,2,2), y=c(1,2,3,4,5,6))

d2 = lazy_dt(d1, immutable = TRUE) %>%
  mutate(
    y=-y,
    z = x+y
    ) %>%
  summarize(
    mean_y = mean(y),
    mean_z = mean(z)
    ) %>%
  mutate(
    sum_mean = mean_y + mean_z
  ) %>%
  as.data.table()


d1
#>    x  y  z
#> 1: 1 -1  0
#> 2: 1 -2 -1
#> 3: 1 -3 -2
#> 4: 2 -4 -2
#> 5: 2 -5 -3
#> 6: 2 -6 -4

Details

Here's my sessionInfo():

R version 4.0.3 (2020-10-10)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19042)

Matrix products: default

locale:
[1] LC_COLLATE=English_Canada.1252  LC_CTYPE=English_Canada.1252    LC_MONETARY=English_Canada.1252 LC_NUMERIC=C                   
[5] LC_TIME=English_Canada.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] dplyr_1.0.3       dtplyr_1.1.0      data.table_1.13.6

loaded via a namespace (and not attached):
 [1] assertthat_0.2.1 crayon_1.4.0     R6_2.5.0         DBI_1.1.1        lifecycle_0.2.0  magrittr_2.0.1   pillar_1.4.7     rlang_0.4.10    
 [9] vctrs_0.3.6      generics_0.1.0   ellipsis_0.3.1   tools_4.0.3      glue_1.4.2       purrr_0.3.4      compiler_4.0.3   pkgconfig_2.0.3 
[17] tidyselect_1.1.0 tibble_3.0.5 

[hadley]

Somewhat more minimal reprex:

library(data.table)
library(dtplyr)
library(dplyr, warn.conflicts = FALSE)

d1 <- data.table(x = c(1,1,1,2,2,2), y = c(1,2,3,4,5,6))

lazy_dt(d1) %>%
  mutate(z1 = 1) %>%
  summarize(z2 = 2) %>%
  mutate(z3 = 4) %>%
  show_query()
#> `_DT1`[, `:=`(z1 = 1)][, .(z2 = 2)][, `:=`(z3 = 4)]

^{Created on 2021-03-02 by the reprex package (v1.0.0)}

Probably something is getting confused here because while the summarise() creates an implicit copy, there's still a mutate before it.