Updated my six year old solution. I thought about doing an __asm__ instruction, but considered the many different architectures the code could run on. Found this compiler intrinsic, supported in GCC and Clang. Other compilers may have their own builtins, but this code only runs on Clang so I didn't bother.

This builtin essentially places the assembly instruction for you, based on the architecture, to have the hardware count the bits in a loop, as opposed to us bitshifting around.

You have a hidden allocation here.

for indx, c := range coord is actually making a copy of coord and iterating it. That's only because you have , c there. If you remove the second for loop value, it will not copy and you will need to index coord yourself at indx.

Nice. Didn't know that was possible. But keep in mind I believe this causes the regex to be compiled on each function call. It would be better to have the regex compiled in a var outside of the function.

Thank you a lot!

ASM learner here: how are you supposed to know which register an argument is passed from C?

Oops, the buffer length on snprintf should only be 25.

Best part is it's ironically over complicated code for just two commands.