_PyUnicode_FromASCII(s, len) is faster than PyUnicode_FromString(s) because PyUnicode_FromString() uses temporary _PyUnicodeWriter to support UTF-8.

But _PyUnicode_FromASCII("hello", strlen("hello"))` is not as easy as `PyUnicode_FromString("hello")`.

_PyUnicode_FROM_ASCII() is simple macro which wraps _PyUnicode_FromASCII which calls strlen() automatically:

```
/* Convenient wrapper for _PyUnicode_FromASCII */
#define _PyUnicode_FROM_ASCII(s) _PyUnicode_FromASCII((s), strlen(s))
```

I believe recent compilers optimize away calls of strlen().