computing ranks in subgroups

Dear R-users!

I am using R 1.0.0 and Windows NT 4.0. I want to compute ranks for the
values of a variable, in several subgroups of a data.frame.
Currently I use:

#example:
x <- data.frame(group=c("A","A","B","A","B","B","A"),

val=c(10,3,9,3,5,1,7))

x

  group val
1     A  10
2     A   3
3     B   9
4     A   3
5     B   5
6     B   1
7     A   7

x.new<-x[order(x$group),  ]
x.new$valgroupranks <- unlist(tapply(x.new$val, x.new$group, rank))
x.new

  group val valgroupranks
1     A  10           4.0
2     A   3           1.5
4     A   3           1.5
7     A   7           3.0
3     B   9           3.0
5     B   5           2.0
6     B   1           1.0

This seems to work, but I wonder if there is a more "natural" (simple?) way
to do that?

by(x[2], x$group, rank)

c(by(x[2], x$group, rank),  recursive=T)