Optimize if_else/coalesce to fifelse/fcoalesce?

[MichaelChirico]

data.table now has its own implementation of if_else (fifelse) and coalesce (fcoalesce).

I just tried the following: add two lines to dplyr/tests/testthat.R:

library(testthat)
library(dplyr)
if_else = data.table::fifelse
coalesce = data.table::fcoalesce

test_check("dplyr")

then run testthat::test_dir('tests').

7 tests failed, but the same tests failed if I removed those two lines, so I'm taking that to mean fifelse and fcoalesce pass usage tests dplyr for behavior users expect of those functions.

Does it make sense/would it be straightforward to optimize those functions when used in dtplyr?

[hadley]

Yeah, I think that makes sense — it's just a matter of adding a few more lines to https://github.com/tidyverse/dtplyr/blob/master/R/tidyeval.R#L61-L75

[hadley]

@MichaelChirico I'm likely to be working on this in the next week if you want to take a stab at it. (If you don't have time, don't worry, as I can also do it myself)

[MichaelChirico]

yes it's still on my to-do :)

…

On Sat, Nov 2, 2019, 10:40 PM Hadley Wickham ***@***.***> wrote: @MichaelChirico <https://github.com/MichaelChirico> I'm likely to be working on this in the next week if you want to take a stab at it. (If you don't have time, don't worry, as I can also do it myself) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#112?email_source=notifications&email_token=AB2BA5I5LNXMEKA43MLDKF3QRWGPJA5CNFSM4I6YT6VKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEC45KKQ#issuecomment-549049642>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2BA5P3IMHBSYMOJBMWBR3QRWGPJANCNFSM4I6YT6VA> .

[MichaelChirico]

Heads up that we've also got fcase in the pipeline, though we opted for a when-value pair API as an initial pass, so it won't be a simple drop-in to the formula interface of case_when:

Rdatatable/data.table#4021