I ran into what I suspect is a rare corner case where .bind_df as used by select_all() will fail if two of the clusters of 100 rows have the same number of columns but different actual columns. In my case, there is a field that is only used in the first 100 rows and a different column that is used only in the subsequent rows. Because the number of colons matches, normal "rbind" is used but it fails.
I ran into what I suspect is a rare corner case where .bind_df as used by select_all() will fail if two of the clusters of 100 rows have the same number of columns but different actual columns. In my case, there is a field that is only used in the first 100 rows and a different column that is used only in the subsequent rows. Because the number of colons matches, normal "rbind" is used but it fails.