select causes type conversion

[dsweber2]

Calling select(epi_data, -col_name) on some epi_df converts it into a tibble. There are probably related issues, since this is only one of many dplyr verbs to support.

[brookslogan]

Can you please provide an example? Sometimes this is intentional: e.g.,

edf %>% select(geo_value)

is not a valid epi df (missing required columns). Sometimes, it's gnarly to fix in general but maybe we could try hacking some fixes for every verb.... e.g.:

edf %>% group_by(geo_value) %>% select(geo_value, time_value, value)

this we probably want to be a grouped epi_df. See probably #223.

[dsweber2]

Apologies if is this is a bit verbose; I was trying to add a temporary column, use it in a calculation, and then remove it.

################### setup###########
n_days <- 40
simple_dates <- seq(as.Date("2012-01-01"), by = "day", length.out = n_days)
rand_vals <- rnorm(n_days)
epi_data <- epiprocess::as_epi_df(rbind(tibble(
  geo_value = "al",
  time_value = simple_dates,
  a = 1:n_days,
  b = rand_vals
), tibble(
  geo_value = "ca",
  time_value = simple_dates,
  a = n_days:1,
  b = rand_vals + 10
)))

epi_data %>% mutate(temp_col = a^2) %>% mutate(perm_col = b-a^2) %>% select(-temp_col)

If we're adding checks as part of extending, insisting that select only changes the non-keys would make sense.