First question:
why copy_noinline ? Its for reducing binary size? I think it may break many optimizations
Second question:
Isn't it more efficient to make memcpy, when possible here ?
Compiler do not optimizes this loop into memcpy:
https://godbolt.org/z/vGPE8jjxT
First question: why copy_noinline ? Its for reducing binary size? I think it may break many optimizations
Second question: Isn't it more efficient to make memcpy, when possible here ? Compiler do not optimizes this loop into memcpy: https://godbolt.org/z/vGPE8jjxT
https://github.com/fmtlib/fmt/blob/4daa3d591f6715671e6716a298ec39f0a34058c5/include/fmt/base.h#L1988