[ARROW-12282] [R] Refactor collect and compute methods

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Component/s: R
• External issue URL: https://github.com/apache/arrow/issues/28089
• Assignee: Neal Richardson
• Reporter: Ian Cook
• Created: 07/Apr/21 20:13
• Updated: 11/Jan/23 08:25
• Resolved: 02/Nov/22 23:16

Description

ARROW-11754 added a compute() method which calls collect(as_data_frame = FALSE). This was a clean and simple way to add compute(), but it would be more logical if collect() instead called compute(), or if both called an internal function that forced computation of the lazy ops. But this is complicated by the need to call restore_dplyr_features() in both compute() and collect() and muddied by some uncertainty about where exactly ensure_group_vars() and ensure_arrange_vars() should be called. Take a closer look at this later date and consider whether to refactor it.

Issue Links

• is fixed by: ARROW-15460 [R] Add as.data.frame.Dataset method