Difference between strncpy and snprintf?

Question

What is the difference between

strncpy(dst, src, n)

and

snprintf(dst, n, "%s", src)

Should I prefer one or the other when trying to do a bounded string copy? And if so, why?

Answer 1

strncpy is not a bounded strcpy. It's a fixed-length (not bounded, fixed-length) operation targeting a destination that's not a C string (null-terminated) but a null-padded field, like what was used in certain kinds of data record tables in the 70s and early 80s. It has basically no modern purpose and should not be used unless you really want this behavior.

snprintf is the canonical C function for "bounded strcpy". On POSIX systems, strnlen+memcpy would be another good alternative, and necessary if you need to support strings longer than INT_MAX. If you don't have strnlen, you can easily make it with memchr and do the same.

Note that the simple use of snprintf is not "bounded" in the sense of having a bound on how much it reads from the source; it necessarily reads the whole source string to determine the return value, which both can incur a time cost and depends on the source being a valid (terminated) string. If you want it bounded in the sense of both source and destination, you can do something like:

snprintf(dest, n, "%.*s", (int)n-1, src);

However, that's sufficiently non-idiomatic and error-prone that I would just use strnlen and memcpy instead.