Accessing repeated bytes field gives a TypeError using jruby

[DanilSerd]

This only happens if the message was encoded and then decoded.

Jruby version: jruby 1.7.25
protobuf version: protobuf-3.0.0-beta-4

This fails:

require 'google/protobuf'

Google::Protobuf::DescriptorPool.generated_pool.build do
  add_message "TestMessage" do
    repeated :repeated_bytes, :bytes, 1
  end

end

TestMessage = Google::Protobuf::DescriptorPool.generated_pool.lookup("TestMessage").msgclass

test_message = TestMessage.new(:repeated_bytes => ["hello".encode("ASCII-8BIT")])
reencoded_message = TestMessage.decode(TestMessage.encode(test_message))

reencoded_message.repeated_bytes[0]

Here's the TypeError:

TypeError: Encoding for bytes fields must be "ASCII-8BIT", but was UTF-8 push at com/google/protobuf/jruby/RubyRepeatedField.java:169 [] at com/google/protobuf/jruby/RubyMessage.java:136 method_missing at com/google/protobuf/jruby/RubyMessage.java:223 (root) at test.rb:19

Failure is on the last line. So it looks like it encodes and decodes the message just fine, but fails on access. Have also tested this with byte map fields, resulting in the same behaviour.

[xfxyjwf]

Does this error also happen if you use the ruby implementation?

[DanilSerd]

@xfxyjwf No, it's not happening with ruby.

[haberman]

I believe this is fixed in #7923.