> rbind_list(data.frame(x=1:3), data.frame(x=NA))
Error: incompatible type (data index: 2, column: 'x', was collecting: integer (dplyr::Collecter_Impl<13>), incompatible with data of type: LGLSXP

This works the other way around.

> rbind_list(data.frame(x=NA), data.frame(x=1:3))
   x
1 NA
2  1
3  2
4  3

Integer promotion to numeric doesn't work either way.

I understand if some strictness is a necessary consequence of a speecy C-implementation. However, if possible, there are at least two harmless promotions that I'd expect to happen automatically.

1. Logical due to all NAs should get promoted to whatever type another piece with actual data has.
2. Integers should get promoted to numeric.