[ARROW-14036] [R] Binding for n_distinct() with no grouping

• Type: New Feature
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 6.0.0
• Component/s: R
• Labels: kernel
• External issue URL: https://github.com/apache/arrow/issues/29634
• Assignee: Percy Camilo Triveño Aucahuasi
• Reporter: Ian Cook
• Created: 18/Sep/21 17:47
• Updated: 11/Jan/23 08:37
• Resolved: 05/Oct/21 14:24

Description

ARROW-13620 added a binding for n_distinct() but it only works for grouped aggregation, not whole-table aggregation. 

This works:

Table$create(starwars) %>%
  group_by(homeworld) %>%
  summarise(n_distinct(species)) %>%
  collect()

but this errors:

Table$create(starwars) %>%
  summarise(n_distinct(species)) %>%
  collect()

#> Error: Key error: No function registered with name: count_distinct

Once we have a non-hash count_distinct aggregate kernel in the C++ library (ARROW-14035) we should bind the options for it in the R package and add a test.

Issue Links

• depends upon: ARROW-14035 [C++][Compute] Implement non-hash count_distinct aggregate kernel
• relates to: ARROW-13620 [R] Binding for n_distinct()