Why is std::dynamic_extent not zero in c++?

Question

For context, I'm trying to make a span type that has a size type parameter, which mimics much of the std::span api.

std::dynamic_extent is not zero in C++, instead being defined as -1ull or std::numeric_limits<std::size_t>::max(). when the Extent ntt parameter of a std::span is std::dynamic_extent, a size member variable of some description is created. But when it isn't, the compile time value for the Extent parameter is used directly for the size instead.

I'm trying to understand why this decision was made, but more importantly, are there any tradeoffs with using 0 instead of max value here?

Answer 1

A known static size of zero is perfectly normal.

While manually deliberately writing it rarely happens (mostly for passing to other, generally generic, code), disallowing it would make generic code more unwieldy.
That's one of the problems with using raw arrays in generic code, aside from them not behaving like normal objects, and probably the main reason std::array is used there instead.

In contrast to zero, which as laid out already serves a necessary role, and in the absence of any negative values in search of meaning, the maximum value is the least needed, and in most cases memory and address space constraints mean it cannot actually occur anyway.