`fill.tbl_lazy()` should set `allow_rename = FALSE` in `eval_select()`

[DavisVaughan]

i.e. right here:

dbplyr/R/verb-fill.R

Line 37 in 860bd6a

cols_to_fill <- tidyselect::eval_select(expr(c(...)), .data)

I'm 99% sure that adding allow_rename = FALSE is correct, as we also do it in tidyr's data frame method.

I think it would have caught some weirdness like this:

library(dbplyr)
library(tidyr) # dev tidyr with PR #1572

df <- data.frame(x = c(1, 2), y = c(2, 2), z = c(1, NA))

df_sqlite <- tbl_lazy(df, con = simulate_sqlite())

df_sqlite |> 
  window_order(x) |> 
  fill(.by = y)
#> <SQL>
#> SELECT `x`, `y`, `z`
#> FROM (
#>   SELECT
#>     `df`.*,
#>     SUM(CASE WHEN ((`.by` IS NULL)) THEN 0 ELSE 1 END) OVER (ORDER BY `x` ROWS UNBOUNDED PRECEDING) AS `..dbplyr_partition_1`
#>   FROM `df`
#> ) AS `q01`

This is a very specific case where in tidyverse/tidyr#1572 we are adding a new .by argument to the generic of fill() that dbplyr doesn't know about yet.

• In the method, .by = y comes through the ... of fill.tbl_lazy() and should get immediately caught by eval_select(allow_rename = FALSE), but isn't
• In the generic, .by = y comes through the .by = argument in the fill() generic, so it happens to not trigger check_dots_unnamed() in the generic