As alexis_laz points out, it has to magrittr:::is_pipedo with explicitly checking for %>%in your expression, and not detecting it and the subsequent logic falling apart %>%because of this.
But why do %>%you need to explicitly search (self or other) %>%in the call?
- %>% , . , %>% , a %>% b %>% c c(b(a)), eval'd ( , b(a) %>% c).
, , - , . .