duplicate colnames gives uninformative error when joining with tibbles

[EdwinTh]

Tibbles don't allow for duplicate column names at creation:

> data_frame(x = 1:2, x = 3:4)
Error: Column `x` must have a unique name

The _join functions throw informative errors when join key is duplicate and x is a regular data.frame

> x <- data.frame(x = 1:2, x = 3:4)
> colnames(x) <- c("x", "x")
> inner_join(x, data.frame(x = 1:2))
Error: Column `x` must have a unique name

However, via the same back door the column names of a tibble can get duplicated. When trying to join then the error is no longer informative.

> x <- data_frame(x = 1:2, x2 = 3:4 )
> colnames(x) <- c("x", "x")
> inner_join(x, data_frame(x = 1:2))
Joining, by = "x"
Error in inner_join_impl(x, y, by$x, by$y, suffix$x, suffix$y, check_na_matches(na_matches)) : 
   Column `` is unknown

If possible, I think the best option here is close the back door route and throw an error at the second line of the last code block. Otherwise an informative error seems desired.

[krlmlr]

Thanks. I think we should check our inputs more stringently, and stop if we encounter a data frame with duplicate names. (We can't entirely prohibit the creation of data frames with duplicate column names.) But this breaking will require some time and planning. In the meantime, I agree that this particular error message should be improved.