scan behaves on JRuby 9.4 differently from MRI

[k77ch7]

Environment Information
jruby -v

jruby 9.4.0.0-SNAPSHOT (3.1.0) 2022-07-11 99fe5ca788 OpenJDK 64-Bit Server VM 11.0.13+8-Ubuntu-0ubuntu1.20.04 on 11.0.13+8-Ubuntu-0ubuntu1.20.04 +jit [x86_64-linux]

Expected Behavior

require 'strscan'

s = StringScanner.new('test string')
s.scan(/\w(\w)(\w*)/) # => "test"
s[0]                  # => "test"
s[1]                  # => "e"
s[2]                  # => "st"
s.scan(/\w+/)         # => nil
s[0]                  # => nil
s[1]                  # => nil
s[2]                  # => nil
s.scan(/\s+/)         # => " "
s[0]                  # => " "
s[1]                  # => nil
s[2]                  # => nil
s.scan(/\w(\w)(\w*)/) # => "string"
s[0]                  # => "string"
s[1]                  # => "t"
s[2]                  # => "ring"

Actual Behavior(jruby 9.4.0.0-SNAPSHOT and strscan-3.0.3-java.gem)

require 'strscan'

s = StringScanner.new('test string')
s.scan(/\w(\w)(\w*)/) # => "test"
s[0]                  # => "test"
s[1]                  # => "e"
s[2]                  # => "st"
s.scan(/\w+/)         # => nil
s[0]                  # => nil
s[1]                  # => nil
s[2]                  # => nil
s.scan(/\s+/)         # => " "
s[0]                  # => " "
s[1]                  # => "s"
s[2]                  # => "tr"
s.scan(/\w(\w)(\w*)/) # => "string"
s[0]                  # => "string"
s[1]                  # => "t"
s[2]                  # => "ring"

I found this behavior by jruby/jruby#7247 .