[R] summarise() returns incorrect results when there are >1 fields to return (Arrow)

[thisisnic]

There's something strange going on with summarise() in Arrow, but not DuckDB, when we return multiple fields.

library(substrait)
library(dplyr)

mtcars %>%
  summarise(mean_hp = mean(hp, na.rm = TRUE), max_hp = max(hp, na.rm = TRUE))
#>    mean_hp max_hp
#> 1 146.6875    335

mtcars %>%
  duckdb_substrait_compiler() %>%
  summarise(mean_hp = mean(hp, na.rm = TRUE), max_hp = max(hp, na.rm = TRUE)) %>%
  collect()
#> # A tibble: 1 × 2
#>   mean_hp max_hp
#>     <dbl>  <dbl>
#> 1    147.    335

mtcars %>%
  arrow_substrait_compiler() %>%
  summarise(mean_hp = mean(hp, na.rm = TRUE), max_hp = max(hp, na.rm = TRUE)) %>%
  collect()
#> # A tibble: 1 × 2
#>   mean_hp max_hp
#>     <dbl>  <dbl>
#> 1    147.   147.

library(substrait)
#> 
#> Attaching package: 'substrait'
#> The following object is masked from 'package:stats':
#> 
#>     filter
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

mtcars %>%
  group_by(am) %>%
  summarise(mean_hp = mean(hp, na.rm = TRUE), max_hp = max(hp, na.rm = TRUE))
#> # A tibble: 2 × 3
#>      am mean_hp max_hp
#>   <dbl>   <dbl>  <dbl>
#> 1     0    160.    245
#> 2     1    127.    335

mtcars %>%
  duckdb_substrait_compiler() %>%
  group_by(am) %>%
  summarise(mean_hp = mean(hp, na.rm = TRUE), max_hp = max(hp, na.rm = TRUE)) %>%
  collect()
#> # A tibble: 2 × 3
#>      am mean_hp max_hp
#>   <dbl>   <dbl>  <dbl>
#> 1     1    127.    335
#> 2     0    160.    245

mtcars %>%
  arrow_substrait_compiler() %>%
  group_by(am) %>%
  summarise(mean_hp = mean(hp, na.rm = TRUE), max_hp = max(hp, na.rm = TRUE)) %>%
  collect()
#> # A tibble: 2 × 3
#>      am mean_hp max_hp
#>   <dbl>   <dbl>  <dbl>
#> 1  127.    127.      1
#> 2  160.    160.      0

^{Created on 2023-02-27 with reprex v2.0.2}