ARROW-14808 [R] Implement bindings for lubridate::date()

[dragosmg]

The following will be supported in arrow (base::as.Date() included to show the difference to date(), but not supported):

library(dplyr)
library(lubridate)
df<- tibble(a= as.POSIXct("2012-03-26 23:12:13", tz="America/New_York"))
df %>%
mutate(
a_date= date(a),
a_date_base= as.Date(a))
#> # A tibble: 1 × 3#> a a_date a_date_base#> <dttm> <date> <date> #> 1 2012-03-26 23:12:13 2012-03-26 2012-03-27

library(arrow)
library(dplyr)
library(lubridate)
df<- tibble(a= as.POSIXct("2012-03-26 23:12:13", tz="America/New_York"))
df %>% arrow_table() %>% mutate(
a_date= date(a)
) %>% collect()
#> # A tibble: 1 × 2#> a a_date #> <dttm> <date> #> 1 2012-03-26 23:12:13 2012-03-26

^{Created on 2022-02-15 by the reprex package (v2.0.1)}

[github-actions]

https://issues.apache.org/jira/browse/ARROW-14808

[jonkeane]

Looking good — a few questions for you

[jonkeane]

One comment + suggestion.

Do we have a ticket to do as.Date separately? Could we do it with this one since it is very very similar?

[jonkeane]

A few comments

[jonkeane]

This is making progress! I didn't quite realize that as.Date() would be as much of a can of worms as it's been — but that's ok, I think it is useful additional functionality!

In the as.Date() implementation it might be nice to restructure the flow so that you have one call at the end to build_expr("cast", intermediate, options = cast_options(to_type = date32())) after having built up intermediate getting things into the right format and then do the cast-to-date (I put intermediate here but that's probably not a good name for it, maybe "just" x like we do elsewhere — I trust you!).

[dragosmg]

I have created an umbrella issue (ARROW-15805) for the possible improvements to the as.Date() binding + ARROW-15800 to implement as_date().

[jonkeane]

This is looking good, a few more questions. I didn't realize just how much extra scope as.Date() would add — though I think this is good functionality to add, so that's ok, but don't feel dispirited this is taking longer!

[jonkeane]

Still getting there, most of the code changes I proposed are attempts to avoid some of the nesting in if/elses. Though I also do think we can support as.Date() with any timezone which will simplify + remove some of the tests down below catching that error.

[dragosmg]

Thanks @jonkeane. Would you mind having another look?

[jonkeane]

This is fantastic, thank you for all of the pushes. This is incredibly close — one minor comment about an additional comment and one suggestion to use compare_dplyr_binding for testing the warning flow "not supported in arrow -> pulling into R". It's functionally the same, but follows the pattern we use elsewhere.

I'm happy to merge even without these, but these two would add a bit of polish + fit

[jonkeane]

Once the CI is green, I'll merge.

Thank you again for all of the work you did on this!

[dragosmg]

Thanks for your patience and support 😉

[ursabot]

Benchmark runs are scheduled for baseline = ce46c1a and contender = 9719eae. 9719eae is a master commit associated with this PR. Results will be available as each benchmark for each run completes.
Conbench compare runs links:
[Finished ⬇️0.0% ⬆️0.0%] ec2-t3-xlarge-us-east-2
[Finished ⬇️0.13% ⬆️0.0%] test-mac-arm
[Failed ⬇️3.21% ⬆️0.0%] ursa-i9-9960x
[Finished ⬇️0.26% ⬆️0.09%] ursa-thinkcentre-m75q
Supported benchmarks:
ec2-t3-xlarge-us-east-2: Supported benchmark langs: Python. Runs only benchmarks with cloud = True
test-mac-arm: Supported benchmark langs: C++, Python, R
ursa-i9-9960x: Supported benchmark langs: Python, R, JavaScript
ursa-thinkcentre-m75q: Supported benchmark langs: C++, Java