Takatoshi Kondo

Which part of msgpack-c are you talking about? C, C++, or both? You wrote `zone::os_default`, so I think your proposal is for C++. Is that right?

Ok. Basically I agree with you. That means I think that we need some custom allocator. Some of embedded environment might not have malloc/realloc/free. NOTE: Due to realloc, that is...

I noticed that some part of msgpack-c core, not including adaptors, uses `std::vector` with default allocator. See https://github.com/msgpack/msgpack-c/blob/master/include/msgpack/v2/parse.hpp#L213 It finally calls malloc. In this case, we can provide the custom...

I don't have much time to implement that big change. If you will do that, I would appreciate! For C case, I think your approach is good. I wrote a...

For C++ case, I don't get the result what is the best approach. Here are some points that I want to share what I thought: 1. `msgpack-c` needs to support...

My C++ concept code is a little bit misleading. I updated it. https://wandbox.org/permlink/VhOqg2zzisic4cB5

Ok, I will review. C++ version first is OK. I will merge only C++ version. But `msgpack::zone` only is not good. I want to merge all C++ custom allocator support...

> Thank you so much for thinking this through with me. I love your analysis of A versus B type allocations. Also, I think your design for zone allocators of...

The core of msgpack-c depends on `char`. For example, https://github.com/msgpack/msgpack-c/blob/master/include/msgpack/v2/parse.hpp#L240 One way to support `std::byte` is replace `xxx` with `basic_xxx` all part of msgpack-c. And provide `typedef basic_xxx xxx;`. Then...

I posted my answer to https://stackoverflow.com/questions/72469007/knowing-end-of-message-in-protocol-with-msgpack/72483972#72483972