The documentation, help("!") e.g.,
contains
> Arguments:
> x, y: logical vectors, or objects which can be coerced to such or
> for which methods have been written.
and then later
> Numeric and complex vectors will be coerced to logical values,
> with zero being false and all non-zero values being true.
and that is consistent with what S-plus (3.4, 5.1, 8.0.2)
gives.
However, (nowadays at least) R internally checks and gives an
error for complex arguments,
even though as.logical(0 + 1i) works fine.
> z <- complex(i=1)
> !z
Error in !z : invalid argument type
> !as.logical(z)
[1] FALSE
I'm proposing to make R's behavior consistent with its
documentation and predecessor.
Martin Maechler
