paradoxical behavior in filter and if_any

[cmorgenstern]

I am trying to find rows via if_any and have run into a very strange behavior. Is this a bug?

my_DF <- data.frame( a = c(1, 1, NA, 1),
b = c(1, NA, 1, 12))
my_DF

my_DF %>%
filter( if_any(everything(), ~ .x == 12 && .y==1))
my_DF %>%
filter( if_any(everything(), ~ .x == 1 && .y==12))

the first filter returns an empty dataframe, which is correct. The second filter causes an error that I can't understand.

Simplifying the code:

my_DF <- data.frame( a = c(1, 1, NA, 1),
b = c(1, NA, 1, 12))
my_DF

my_DF %>%
filter( if_any(everything(), ~ .x == 1 && .y==12))

Still produces the error.

I am new to the syntax involving if_any so maybe I am making some sort of beginners error?

Any help appreciated!

Carl

[nirgrahamuk]

You should say what your intent is / what you believe the behaviour should be ...

It seems like you think you should inherently have access to analyse pairs of columns, (.x .y ) being shortcuts to them... but you've not set up anything like that.
my_df happens to have 2 columns, but it might have had any number, 1,2,3....
everything() would mean you are processing those 'any number of columns' and you can apply a test on the contents of them (.x == etc.) individually, then if_any will then be true if any of the columns test true. effectively combining them at that level for the filter.