dplyr::case_when() combined with mutate() does not work when applied to tbl

[CarolineBarret]

I am working with R 3.4.3 and dplyr 0.7.4.

I am trying to apply the case_when() function to a tibble object from a database. But when I combine the case_when() function to the mutate() function, I get an error:

dir.create("tmp")
# The datasets.sqlite is just a sqlite database 
# It contains a table called iris with the iris data frame in it
file.copy("data/datasets.sqlite", "tmp")

# Connect to the database
con <- DBI::dbConnect(RSQLite::SQLite(), "tmp/datasets.sqlite")

cake <- dplyr::mutate(dplyr::tbl(con, "iris"),
                        type = dplyr::case_when(
        Sepal.Length > 5.8 | Petal.Length > 3.7 ~ "long",
        Sepal.Width > 3  | Petal.Width > 1.2 ~ "wide",
        TRUE ~  "other"
      )
    )

This gives me the following error: Error in eval_bare(f[[2]], env) : object 'Sepal.Length' not found

I think this is to do with the fact that I use mutate() because I do not get an error with the following:

  ir <- dplyr::tbl(con, "iris")
  cake <- dplyr::case_when(
                          ir$Sepal.Length > 5.8 | ir$Petal.Length > 3.7 ~ "long",
                          ir$Sepal.Width > 3  | ir$Petal.Width > 1.2 ~ "wide",
                          TRUE ~  "other"
                        )

It also is to do with tibbles output (and not with data.frames), because this works fine too:

  cake <- dplyr::mutate(DBI::dbReadTable(con, "iris"),
                        type = dplyr::case_when(
                          Sepal.Length > 5.8 | Petal.Length > 3.7 ~ "long",
                          Sepal.Width > 3  | Petal.Width > 1.2 ~ "wide",
                          TRUE ~  "other"
                        )
  )

[krlmlr]

Thanks, this seems related to the database backend only.

/move to dbplyr

[ghost]

This issue was moved by krlmlr to tidyverse/dbplyr#135.

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/