On Thursday, 1 November 2012 at 15:23:35 UTC, David Nadlinger wrote:
You avoid having to flip the byte order at the application level on the prevalent (at least in the domain where protobuf is used) architectures that way.

Yes for platform native numbers, but not for ULEB - which is platform-independent and is read by the same (source) code on all platforms AFAIK. I see no gain for little endian platform here: registers work the same on BE and LE platforms. One could even say ULEB is intentionally incompatible with any platform so that you don't mess things up in homogeneous environment writing native ints into stream (which will break in heterogeneous environment). This was a design choice for exi to use ULEB for strings too instead of utf8.

Reply via email to