Personally, I'd find it useful if gfirst()/glast() would accept a na.rm argument. This would then return the first/last non missing element of a vector (within each group).
For consistency I guess also first()/last() should then accept a na.rm argument. Although I'm not sure what this would mean if a non-atomic object (i.e. a data.frame) was passed. Maybe the flag should only be honored if is.atomic(x) && nargs() == 1L?
Personally, I'd find it useful if
gfirst()/glast()would accept ana.rmargument. This would then return the first/last non missing element of a vector (within each group).For consistency I guess also
first()/last()should then accept ana.rmargument. Although I'm not sure what this would mean if a non-atomic object (i.e. adata.frame) was passed. Maybe the flag should only be honored ifis.atomic(x) && nargs() == 1L?