It does seem like the values CLOCKS_PER_SEC, _PyTime_MAX and SEC_TO_NS are all defined at compile-time, so presumably this can be a compile-time error.

As currently defined, though, it seems very unlikely that this would ever be false, since _PyTime_MAX is the same as LLONG_MAX, which cplusplusreference says is `9223372036854775807 (2**63-1) or greater` [0], so CLOCKS_PER_SEC would need to be >= 2**50, which is unlikely to be... accurate.

In any case, it seems like this check can happen at compile-time, either resulting in compilation failing, or just undefining HAVE_CLOCK.

[0] http://www.cplusplus.com/reference/climits/