Unable to use names_glue within pivot_wider() when NA is the only unique value

[camnesia]

Unable to use names_glue within pivot_wider() when the only unique value is NA - the same code does not cause any problems when lazy_dt() is not used.

example <- fish_encounters %>%
  mutate(test = NA_integer_) %>%
  lazy_dt() %>%
  pivot_wider(., 
              names_glue = "{test}",
              names_from = test, 
              values_from = seen,
              names_repair = 'minimal') %>%
  collect()

names_glue does work, if there are other values available in names (for example using the station variable). Using a variable with only NA does seem to be possible but only if names_glue is not used at all.

example <- fish_encounters %>%
  mutate(test = NA_integer_) %>%
  lazy_dt() %>%
  pivot_wider(., 
              names_glue = "{station}",
              names_from = station, 
              values_from = seen,
              names_repair = 'minimal') %>%
  collect()

[markfairbanks]

Thanks for catching this - we'll take a look.

Smaller reprex:

library(dtplyr)
library(dplyr)
library(tidyr)

df <- tibble(x = c("a", "a"), y = c("a", NA), z = 1:2)

df %>%
  lazy_dt() %>%
  pivot_wider(names_from = y,
              values_from = z,
              names_glue = "{y}",
              names_repair = "minimal") %>%
  collect()
#> Error: NA in 'new' at positions [1]