If you are seeing this error in code that used to work, the most likely cause is a change dbplyr 1.4.0.

General
1

Hi.

A quick question. I have updated shiny app from 3.5.1 to 3.6.1 with all the packages. The code now breaks becuase of new version of dplyr/dbplyr with fairly self-explanatory error message If you are seeing this error in code that used to work, the most likely cause is a change dbplyr 1.4.0. Previously df$xordf[[y]]implied thatdfwas a local variable, but now you must make that explict with!!orlocal(), e.g., !!df$x or local(df[["y"])) .
Quick googling shows that it affects %in% function. Are there other functions where I could expect problems?

Is there a way to check where exactly it breaks?
Thanks!

2

What does change is code like that:

some_vector <- list(some_values = c("value_1", "value_2"))
df %>%
  dplyr::filter(x %in% some_vector$some_values)

If your df is a remote database connection, then dbplyr will assume that some_vector is a column in your remote dataset and will try to look for some_values in it. To avoid that, you'll want to do something like what is suggested in the error message:

some_vector <- list(some_values = c("value_1", "value_2"))
df %>%
  dplyr::filter(x %in% !!some_vector$some_values)

# or

some_vector <- list(some_values = c("value_1", "value_2"))
df %>%
  dplyr::filter(x %in% local(some_vector$some_values))

The reason this is happening is because most modern DB's have some support for json and now some_vector$some_values can be used to extract some_values element from some_vector in your DB. Therefore, you need to explicitly tell dbplyr where to look for some_vector$some_values

3

Thanks Misha. I will consider rollback of dbplyr, cause it keeps jumping at me even though it should not at the first glance. The app reactivity inheritance between modules is fairly complex.