Why is MsgPack big endian?

[DaveVdE]

Not really an issue per se, but I'm writing a small encoder/decoder and I must say, it's a pain having to tansform all my in-memory representations of ints, floats from little-endian to big-endian.

Is it deliberate? Is it an optimization for smaller (possibly big-endian) mobile platforms? I care to know.

[frsyuki]

It's because a lot of standard network protocols use big-endian. "Network byte order" is big-endian. My design choice was to follow the traditional convention because I hadn't believe which endian becomes the majority when I designed the format. One practical benefit is that we can use portable and fast macros such as htons(3) or htonl(3) to convert host byte order to network byte order.

[DaveVdE]

Thank you.

[Tronic]

Sorry to post to a very old thread, but just for the public record:

Any new formats should really be little endian. The network convention is dead and no CPUs use big endian anymore. Little endian buffers work directly on current CPUs and are much easier to handle whenever a byte at a time needs processing (avoids some backwards for loops that you need with big endian).