If I have pointers of type T (T*) and I have an array of them T* array[N] will these two methods allow me to later check which entries are null pointers to lazily initialize stuff in each bucket?
memset(array, 0, sizeof(T*)*N);
or
for (int i = 0; i < N; ++i)
array[i] = NULL;
i.e. will the memset call also let me later do if (array[i] == NULL) ...?
I wouldn't want to introduce undefined behavior if not..
Although a null pointer value isn't technically required be all zero bits I'm not aware of any system where it's not all zero bits. So either method should work.
However there's an easier way to initialize an array which will set the correct null pointer values even on some evil implementation:
T *array[N] = {};
Or if you dynamically allocate it:
T **array = new T*[N]();
Formally, the memset approach doesn't work, because there is no requirement that a null pointer be represented by the value 0 in memory. In practice it works fine.
Better than both: std::uninitialized_fill; the standard library implementor can do things that you can't to optimize performance.
std::uninitialized_fill.
std::uninitialized_fill over std::fill?std::uninitialized_fill is more pedantically correct even if unnecessary in this case.
std::uninitialized_fill and std::uninitialized_copy use placement new to construct the elements they are initializing. std::fill and std::copy use assignment to already constructed objects. The difference is irrelevant to trivially constructible types such as pointers.
T* array[N] = {NULL};