Just a note: defining own integer types has sense for resource-limited platforms. Most common type I see is something like "dim_t", which is 32-bit or 64-bit de

Just a note: defining own integer types has sense for resource-limited platforms. Most common type I see is something like "dim_t", which is 32-bit or 64-bit depending on use-case. 32-bit integers are often used even on 64-bit platforms in pointer compression schemes (for example, allocate your own heap and only store 32-bit offsets). This not only gives 2x improvement on memory usage for <4GB workloads, but it also improves performance due to better cache locality.