.sort.uniq or .uniq.sort

Which is better, .uniq.sort or .sort.uniq, to sort and make the array unique?

I thought it would be wrapped around a long one, but when I executed it destructively, it turned out that there were cases where .uniq! .Sort! Could not be executed @ 2.7.1

test.rb

r = {
  's' => [5,1,3,53,2],
  't' => [5,3,4,4,3,2],
}
r.each_key{|k|
  #r[k].sort!.uniq!
  r[k].uniq!.sort!
  #  =>No good
}
p r

python

$ ruby test.rb
Traceback (most recent call last):
	2: from hoge2.rb:5:in `<main>'
	1: from hoge2.rb:5:in `each_key'
test.rb:7:in `block in <main>': undefined method `sort!' for nil:NilClass (NoMethodError)

You can do it with .sort! .uniq!.

Normally,

test2.rb

a = [5,3,4,4,3,2]
#a.sort!.uniq!
a.uniq!.sort!

p a

If so, there is no error.

Even if the bench is slow, I wonder if I will use .sort.uniq [^ 1]

[^ 1]: This one is more familiar with the shell.