When filter() gets a formula object, it modifies it by reference

[lionel-]

This happens when we forget to add an underscore to use the non-NSE version:

my_filter <- ~ am == 1
mtcars %>% filter(my_filter)
## Error: filter condition does not evaluate to a logical vector. 

Now, my_filter contains the actual am object instead of just its name.

[astamm]

I think you have to use the SE version filter_ of filter in this case.

On Feb 16, 2015, at 6:06 AM, Lionel Henry <notifications@github.commailto:notifications@github.com> wrote:

This happens when we forget to add an underscore to use the non-NSE version:

my_filter <- ~ am == 1
mtcars %>% filter(my_filter)

Error: filter condition does not evaluate to a logical vector.

Now, my_filter contains the actual am object instead of just its name.

—
Reply to this email directly or view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_hadley_dplyr_issues_971&d=BQMCaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=YY0Nl6LrGyElaL-CxD3xdZzVapE6UH4Cpr-mdEV4EX3NxhgDF5KLPTqjD9TVadjr&m=-mJH4C0D9D9xKxrIZiTnmFCsYNMArBOdNjMJR4gXmMQ&s=eZOapKcu4RIYMaB4wrCogZdLkopLAb2IZJ2RC93fZgA&e=.

[lionel-]

Sure, but this is no reason for filter() to mess with my formula ;-)

[hadley]

This isn't a crasher, so putting off until 0.5

[romainfrancois]

Hmm. I don't get it, since we do duplicate it in the Lazy class:

        inline SEXP expr() const {
            return Rf_duplicate(data[0]) ;
        }