I am using stb image library to load a image into C-style contiguous array. I want to move this array into a xtensor style xtensor or xarray with xt::adapt. Here is the demonstration code:
// It must be C-stype array because stb is C library.
std::size_t size = 24;
unsigned char arr_1d[size];
for (auto i = 0; i < size; ++i) {
arr_1d[i] = i + 1;
}
// Convert 1D array with 24 elements into (3, 2, 4) xtensor.
auto img = xt::adapt(arr_1d, xt::xshape<3, 2, 4>(), xt::no_ownership());
But I kept getting the error:
test_xtensor.cpp:16:14: error: no matching function for call to 'adapt'
auto img = xt::adapt(arr_1d, xt::xshape<3, 2, 4>(), xt::no_ownership());
^~~~~~~~~
/tmp/xtensor-install/include/xtensor/xadapt.hpp:85:5: note: candidate template ignored: substitution failure [with L = xt::layout_type::row_major]: variably modified type 'unsigned char (&)[size]' cannot be used as a template argument
adapt(C&& container, const SC& shape, layout_type l = L)
Is it possible to achieve the above task with C-style array or I must convert to C++ vector first?
The following compiles:
auto img = xt::adapt((char *)arr_1d, xt::xshape<3, 2, 4>());
As I understand the documentation, it ought to work without the cast, but I get the following error about that candidate not being considered: (Clang 14 on macOS)
/opt/homebrew/include/xtensor/xadapt.hpp:458:17: note: candidate template ignored: could not match 'T[N]' against 'unsigned char[size]'
inline auto adapt(T (&c_array)[N], const SC& shape, layout_type l = L)