tidyr::fill(everything()) fails for character columns

[lindeloev]

This fails and I would expect it to fill the NA in the character column:

data.frame(x = c("A", NA, "B")) %>%
    dtplyr::lazy_dt() %>%
    tidyr::fill(everything())

Error in nafill(x, "locf") :
'x' argument must be numeric type, or list/data.table of numeric types

It runs as expected without everything() and on numeric data, but the mere presence of a character column causes it to fail.

R 4.1.1
dtplyr 1.2.0 installed from CRAN
tidyr 1.1.4 installed from CRAN

[markfairbanks]

Unfortunately this is a limit of data.table as it doesn't support filling with character columns yet (see Rdatatable/data.table#3992).

For now we would recommend using tidyr::fill() directly on a data frame instead of a lazy_dt.

[markfairbanks]

In the documentation you'll see it mentioned that fill() only works on numeric columns in dtplyr.

There was a design discussion here around how to handle fill() in dtplyr if you want some more context.