not sure if this is intentional, possible improvement to rbind.fill()
rbind( NULL , mtcars )
rbind.fill( data.frame() , mtcars )
rbind.fill( NULL , mtcars )
This used to work at least in 1.7.1.
It seems that the fix to issue #51 is at fault.
Do you have a real use case where this causes a problem?
Yes, I'd say so. Something like
dfs = lapply(...)
df = do.call(rbind.fill, dfs)
Depending on what the function under lapply does and what kinds of data it happens to have it's fully possible to get a NULL value instead of a zero-length data frame. And since at least unlist and rbind gracefully discard NULLs, I'm not used to trying to avoid them.
Makes sense - now fixed :)