Takao Nakaguchi

Thanks for your comment, oza. The exception you pointed seems to be designed to be thrown in test code. See the test code of testNormal03 method. There is the comments:...

Thank you for the consideration. I investigated fluentd feature, it seems to be designed to accept JSON map (not JSON array or values). So at least we should limit data...

writeObj method writes map value of POJO. Neither @Message nor @Beans can write map value, these write array value. When you execute following code, ``` java System.out.println(mp.read(mp.write(new Msg("bob", 20)))); ```...

> Your proposal sends "strForTesting" => "hoge" as a part of events when Msg objects are serialized. Is it really a > part of you'd like to send? Yes. Converting...

I implemented @Message(style=SerializationStyle.MAP). Here is the difference to original msgpack-java: https://github.com/takawitter/msgpack-java/compare/mapstyle What do you think of this kind of change? I think that is not attractive for msgpack-java comitters because...

I added the commit that realizes above a. and b. But not yet c. Because I don't have good way to exposure Msgpack instance RawSocketSender has. best regards.

I'm not sure, but using ByteBuffer may cause serious performance decrease because of copy operation from native memory to heap. But, anyway, you can try and measure performance. I recommend...

I see. FYI, according to the URL below, using Unsafe class and DirectBuffer class seem to provide faster access to natively allocated memory like this: ``` java Field field =...

That's so nice!!! I can't wait your pull-req (Do you plan to do that?) Ahh.. Yes, indeed, the tail array is a memory eater. Though I want to improve that,...

That's great! I'm looking forward to your pull req :) I read Japanese book http://www.amazon.co.jp/gp/product/4774149934/ (Technologies to support Japanese input method editors) and also gather information from web (all Japanese):...