Ok, I think I understood the issue :-) The problem is when the uint32_t type is present but is not exactly 32-bit width. When using uint32_t, *I* expect that an array of uint32_t items will takes 4 x n bytes. In which case is it interesting to use an uint32_t which may be bigger? If there is an use case (speed maybe?), we should define a Py_uint32_t which is always present and always exaclty 32 bits. If there is no use case (ex: int_fast32_t or int_least32_t can be used instead), it is maybe better to replace the available type using a #define to use the expect type.