Repository navigation
[FEAT] Support for per-operation grouping within fsummarise/fmutate akin to dplyr's .by argument #881
Description
What aspect of the package is your request related to?
dplyr introduced per-operation grouping support via a .by argument within functions like summarise and mutate in 2023 (dplyr 1.1.0), thereby avoiding explicit group_by() and ungroup() calls in a pipeline. A similar option for fsummarise and fmutate would be great!
Describe the solution you'd like
Instead of
<some pipeline> %>%
fgroup_by(<group vars>) %>%
fsummarise(<summary vars>) %>%
fungroup() %>%
<more pipeline>
allow for
<some pipeline> %>%
fsummarise(
.by = <group vars>,
<summary vars>
) %>%
<more pipeline>
and similarly for fmutate.
Additional context
This would allow for a virtual drop-in replacement of summarise or mutate functions into existing code, with much smaller pipeline modifications.
Activity
SebKrantz commentedon Sep 12, 2026
Hi @wgrundlingh. Thanks! I am not currently planning to implement this, as fgroup_by() grouping in collapse is really efficient, there are g arguments to the fast statistical functions which can be used, especially in fmutate() calls, and there is a convenience function for aggregation called collap(). Plese carefully read this vignette for recommended programming patterns and also see my answer to #674. I also note that the Claude Code Skill now provides expert guidance on collapse to agents.