Error with summarise in Arrow

[thisisnic]

library(arrow)
library(substrait)
library(dplyr)

# works in dplyr
mtcars %>%
  select(mpg, cyl, vs, am, gear, hp) %>%
  group_by(am) %>% summarise(s = sum(hp)) %>%
  collect()
#> # A tibble: 2 × 2
#>      am     s
#>   <dbl> <dbl>
#> 1     0  3045
#> 2     1  1649

# works in duckdb
mtcars %>%
  duckdb_substrait_compiler() %>%
  select(mpg, cyl, vs, am, gear, hp) %>%
  group_by(am) %>% summarise(s = sum(hp)) %>%
  collect()
#> # A tibble: 2 × 2
#>      am     s
#>   <dbl> <dbl>
#> 1     1  1649
#> 2     0  3045

# doesn't work in arrow
mtcars %>%
  arrow_substrait_compiler() %>%
  select(mpg, cyl, vs, am, gear, hp) %>%
  group_by(am) %>% summarise(s = sum(hp)) %>%
  collect()
#> Error in sum(hp): invalid 'type' (list) of argument

[paleolimbot]

That's because there are no aggregation functions registered for arrow! It's basically trying to do:

base::sum(substrait:::simple_integer_field_reference(0))
#> Error in base::sum(substrait:::simple_integer_field_reference(0)): invalid 'type' (list) of argument

^{Created on 2022-10-19 by the reprex package (v2.0.1)}

...because sum() isn't defined in arrow_funs (but it is in duckdb_funs).

[thisisnic]

Aha, thanks!