as.roman(c('1', '2', '3'))
 I II III
as.roman(c('1', '2', '3', ''))
 <NA> <NA> <NA> <NA>
In .roman2numeric(x) : invalid roman numerals: 1 2 3
I'm assuming this is unexpected behaviour and the expected output should be:
 I II III <NA>
The source code currently has a
## (hopefully, for the time being)
comment, i.e., a hidden "FIXME".
I will commit a bug fix which will also contain some new features for as.roman() and the "roman" class probably later today.