Expressive and elegant. And it compiles to just 4 instructions (for x64 and GCC at least).

Well, if we're trying to learn how to efficiently solve actual day-to-day programmer problems, this is the best solution I'd imagine.

For experienced C++ coders, is this a clear and easily grokable solution?

I learn new standard library methods by seeing other people's solutions WAY faster than reading by the references.

Is using using t better than simply using your int i?

Nice! This is the fastest implementation based on my rough profile testing. Efficient!

This is a quite different language from C...

Based on my understanding, it is a difference of stack variable vs heap variable.

char res[len + 1] would make a local variable on the stack.

There are two problems with this. 1) it is a local variable, so it can't be easily seen or used outside of the function. 2) it is on the stack, and the compiler needs to know ahead of time exactly how much space to put on the stack.

If you don't know ahead of time how much space you need, the variable must be allocated on the heap. char *res = malloc(len + 1); creates a space on the heap.

Since we don't know ahead of time how long the input DNA sequence is going to be, the program will only work if you allocate it on the heap.

Also, the test suite that "uses" our code expects that we allocated on the heap, so it will try to free().

You don't need to check == 1 because 1 is already true.

Nice doing the calculatuon with only 1 loop.

This was a fantastically fun kata to do--got to delve into the weeds of assembly, registers, and stack address calculations! I kept tripping over C statements that cobblered registers. This was a joy of a kata.