Improve error message for joining ordered factors?

[gklarenberg]

Would it be possible to improve the error message that is displayed when one is trying to join ordered factors (ordinal data) with different levels?

> df1 <- data.frame(D.1 = c(1,3,4,2,5),
+                  D.2 = factor(c(4,5,3,1,2), ordered = TRUE),
+                  F.4 = c(5,4,3,2,1))
> df1$D.2
[1] 4 5 3 1 2
Levels: 1 < 2 < 3 < 4 < 5
> 
> df2 <- data.frame(D.1 = c(6,2,3,4,1),
+                   D.2 = factor(c(NA,3,3,5,1), 
+                                exclude = NULL,
+                                ordered = TRUE,
+                                levels = c(1,2,3,4,5,NA)))
> df2$D.2
[1] <NA> 3    3    5    1   
Levels: 1 < 2 < 3 < 4 < 5 < <NA>
> 
> df3 <- full_join(df1, df2)
Joining with `by = join_by(D.1, D.2)`
Error in `full_join()`:
! Can't join `x$D.2` with `y$D.2` due to incompatible types.
ℹ `x$D.2` is a <ordered<cd34a>>.
ℹ `y$D.2` is a <ordered<88c4b>>.
Run `rlang::last_trace()` to see where the error occurred.

Granted - I am not sure how common this is, but I ran into it when trying to join survey data (obtained with the qualtRics package by @juliasilge), where NA was added to the levels (when participants skip a question). The error message was really difficult to decipher ("incompatible types", just "types"?), I could not find anything online. Only when I started making a reprex for a qualtRics issue, I realized what the problem was.

[DavisVaughan]

Tracking at r-lib/vctrs#1859