ianmcook commented on a change in pull request #11018:
URL: https://github.com/apache/arrow/pull/11018#discussion_r711111279
##########
File path: r/R/dplyr-summarize.R
##########
@@ -136,6 +136,17 @@ summarize_eval <- function(name, quosure, ctx, recurse =
FALSE) {
return()
}
+ if ("median" %in% funs_in_expr) {
+ expr <- wrap_median(expr)
Review comment:
@nealrichardson this implementation will not actually work until #11159
is merged and we have a `list_element` scalar kernel, but the gist of this as
follows:
We need to translate `median(x, na.rm)` to `list_element(tdigest(x, options
= list(q = 0.5, skip_nulls = na.rm)), 0)` but because of how aggregate function
bindings work, it isn't possible to do that directly in the definition for
`agg_funcs$median`, so I hacked together this solution in which
`summarize_eval()` calls a function named `wrap_median()` that recurses through
the expression AST and replaces `median(...)` with
`arrow_list_element(median(...), 0)`.
Can you think of a more straightforward immediate solution to this?
Unfortunately `substitute()` won't work here; I tried that before resorting to
the AST recursion stuff.
I think we should ask for `median` and `hash_median` kernels in the C++
library, to avoid having to do all these contortions in the R bindings.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
